Noninvasive techniques for detection of foreign

0 downloads 0 Views 2MB Size Report
The performances and limitations from previous studies in several types of food appli- ... mentation of the noninvasive technique as a monitoring tool have received an encouraging ..... cross validation detection rate achieved were 100, 98.5, and 93.5% for large .... The size range of the foreign bodies was from 1 to 3 mm.
Received: 9 November 2017

|

Revised: 19 March 2018

|

Accepted: 4 April 2018

DOI: 10.1111/jfpe.12808

REVIEW ARTICLE

Noninvasive techniques for detection of foreign bodies in food: A review Mohd Taufiq Mohd Khairi

|

Sallehuddin Ibrahim | Mohd Amri Md Yunus |

Mahdi Faramarzi Department of Control and Mechatronics Engineering, Faculty of Electrical Engineering, Universiti Teknologi Malaysia, Skudai, Johor 81310, Malaysia Correspondence Sallehuddin Ibrahim, Department of Control and Mechatronics, Engineering, Faculty of Electrical, Engineering, Universiti Teknologi Malaysia, Skudai, Johor 81310, Malaysia. Email: [email protected]

Abstract Foreign body in food is a matter of concern in food industry as it determines the safety and quality of the product. This issue leads to anxiety as it is hazardous if it is accidentally being consumed. Therefore, foreign body detection is urgently needed. Noninvasive techniques are attractive as they can be used to evaluate without altering the originality of the food products in terms of ingredient and structure. With the growing interest in this subject, this article reviews the development of noninvasive techniques including X-ray, thermal imaging, near-infrared spectroscopy, hyperspectral imaging, ultrasonic, and terahertz. The principle and application of each technique are elaborated. The performances and limitations from previous studies in several types of food appli-

Funding information Ministry of Higher Education Malaysia (MyBrain15 program); Universiti Teknologi Malaysia, Grant/Award Numbers: 15H85 and 4J255

cations are analyzed. In addition, future trends and challenges encountered with these techniques are highlighted. It is envisaged that the information gathered in this article will be a valuable source of information for researchers working in this topic.

Practical applications The existence of foreign bodies in food causes degradation of their quality and safety. The implementation of the noninvasive technique as a monitoring tool have received an encouraging response from the manufacturers which endeavors to maintain their credibility and reputation. This review serves to provide an up-to-date development on the noninvasive technique for detecting foreign bodies which is useful for academicians and researchers especially from the food processing industries.

1 | INTRODUCTION

glass, metal, or rubber. The risk level foreign body depends on the size, type, hardness, and sharpness of the object. The existence of foreign

Foreign body is referred to as any extraneous object or foreign matter

body in food can cause choking when it is eaten. There are also cases

in food item which may cause illness or complications to a person at

where surgery was required to remove the foreign body as reported in

the time of consumption. Foreign body might accidently enter food

(Bansal, Singh, Mangal, Mangal, & Kumar, 2017; Hyman, Klontz, &

due to unsanitary conditions during production, processing, handling,

Tollefson, 1993; Olsen, 1998). Such cases may result in the food

storage and distribution of food. Foreign bodies’ sources are catego-

manufacturer losing customers’ trust and incurring significant losses.

rized into two types; intrinsic and extrinsic (Edwards & Stringer, 2007).

Food manufacturers will perform several investigations at every

Intrinsic type refers to the detected foreign bodies which are related to

line of operation, starting from the collection of the raw materials until

the food and resulted from the process of preparing foods. Examples

the delivery process to determine whether it is a foreign body or not

of these might include fruit stalks in dried fruit and bone fragment in a

(Lewis, 1993; Marsh & Angold, 2004). From the results, they will esti-

meat product. Whereas for extrinsic types, the foreign bodies that exist

mate the types of materials that may be produced together with the

are not directly related to the food product such as the presence of Two potential reviewers of the manuscript (1) Dr. Jayani Chandrapala (Email: [email protected]) and (2) Prof. Dr. Michael Ngadi (Email: michael. [email protected]).

J Food Process Eng. 2018;e12808. https://doi.org/10.1111/jfpe.12808

food and will ensure the food is safe from contamination. Other than that, they also refer to the records of customers’ complaints to determine whether it is foreign body or not. However, they need to be careful regarding the customer complaints because sometimes the alleged

wileyonlinelibrary.com/journal/jfpe

C 2018 Wiley Periodicals, Inc. V

|

1 of 20

2 of 20

|

MOHD KHAIRI

ET AL.

foreign body is impossible to be found in the type of food that they

Smith, and Batchelor (1998), Edwards (2004), and Peariso (2006).

produced. Mostly, the case is due to sabotage by the employees of the

Although review work has previously been done, many novels noninva-

company and the consumers themselves (James, 2005; Marsh &

sive techniques have come into sight in recent years. Therefore, it

Angold, 2004). Sabotage by employees is difficult to be detected

becomes an impetus for this review. In this article, a review on the lat-

because it can be carried out in any phase starting from the production

est techniques and approaches has been conducted to give a clear

phase until the delivery phase. The contamination which originated

viewpoint on applications for detecting the foreign bodies. The funda-

from the consumers may occur, for example when they opened the

mental principles of the technique including X-ray, thermal imaging,

cap of a can using a can opener. The can fragment may accidently enter

near-infrared spectroscopy, hyperspectral imaging, ultrasound, and ter-

the food when they pull and rotate the can opener.

ahertz are briefly elaborated. Various research regarding those tech-

Foreign bodies like stone and insects may come from unsanitized

nique for detecting foreign bodies in several types of foods are also

factory environment (Woh, Thong, Behnke, Lewis, & Mohd Zain,

analyzed and discussed. The advantages and disadvantages for each

2016). It also can exist in each operating unit such as in processing,

technique are also expounded in this article.

handling, distribution, storage and delivery unit (Stringer & Hall, 2007). In the processing and handling unit, the objects that could be present

2 | NONINVASIVE TECHNIQUES

are glass, nuts, bolts, grease, and hair. The failure of machine system and the employees’ negligence are the biggest contributors to the exis-

The various densities of the foreign bodies provide a challenge to the

tence of such object. The process of early detection and removal of the

sensor’s ability (Mohammadi, Ghasemi-Varnamkhasti, Ebrahimi, &

object is vital before the product is moved to the next processing stage.

, Ch Abbasvali, 2014; Salazar, Turo avez, & García, 2004). High density

Early removal is very important to ensure that the machinery and the

materials such as metal and stone are more easily distinguished by

equipment in the next production line are not damaged due to these

most of the sensing techniques. However, not all conventional techni-

objects. Objects like metal and stone fragment may cause damage to

ques can detect low density materials such as insect and plastic (Li, Liu,

the cutting or chopping tools as they are cut together with the raw

Sun, Ma, & Ding, 2015). The detection process is easier in the case of

food such as fruit or meat (Stringer & Hall, 2007; Trafialek, Kaczmarek,

foods that have divergent phases compared to the foreign body. A sim-

& Kolanowski, 2016). For example, the metal blade of a food process-

ple case for example is the presence of foreign body in dairy or bever-

ing system may cracks resulting in small fragments of the blade to

age product. The reliabilities of these technique sometimes are limited

break off into the food. In the distribution and storage process, there is

if the type of food and foreign bodies are of the same phases. For

a risk when a food container is made of glass. The container can crack

example, in the case of stone mixed with cereals, both stone and cere-

when it hits hard object like pallets and the glass fragments unexpect-

als are solid. It is difficult to detect as some imaging sensing techniques

edly mixed with the food (Edwards & Stringer, 2007). Besides, the food

have problems in distinguishing between the stone and the cereals.

products should be kept and placed in a safe and clean environment as

However, the detection process may be easier if the shape and size of

well as maintained at a suitable temperature to ensure insects or pests

the stone are significantly different compared to the cereals. Thus,

are not attracted. In the delivery unit, the transportation storage sys-

numerous initiatives and innovations have been carried out to improve

tem plays a vital role in ensuring that the product is distributed safely

the reliability and capability of the sensors for such cases.

to the retailers and shops. Several issues of concern such as the storage temperature level for the transportation, cleanliness and time span for delivering the product. This precautionary measure may minimize the probability of the food product from being contaminated.

2.1 | X-ray X-ray is a form of electromagnetic radiation having wavelengths and

Early detection of foreign bodies is an important control measure

photon energies in the range of 0.01 to 10 nm and 120 eV to 120 keV,

in ensuring the safety and quality of food product. The food industry

respectively (Mathiassen, Misimi, Bondø, Veliyulin, & Østvik, 2011). X-

makes numerous efforts to avoid unwanted foreign objects in food.

ray is categorized as “soft X-ray” when the photon energy is up to

The conventional methods that have been used to detect foreign

about 10 keV (10–0.10 nm wavelength), whereas those photon ener-

bodies include: metal detector, magnet, electrical impedance, and sur-

gies within the range from 10 to 120 keV (0.10–0.01 nm wavelength)

face penetrating radar. Although these methods are simple and reliable,

are classified as “hard X-rays” (Chen, Zhang, Zhao, & Ouyang, 2013). X-

their weaknesses have restricted their applications. Nonconductive

ray can be used to identify the density characteristics of a specific

materials such as plastic and glass cannot be detected using metal and

object/material and provide valuable information on any nonuniformity.

magnet detection system. Electrical impedance and surface penetrating

The system captures a greyscale image which reveals the density com-

radar do not work well with metallic or foil packaging. The noninvasive

position of the object. The compositional differences come from the

technique has gained attention as the evaluation process can be per-

variation of X-ray attenuation which includes absorption and scattering

formed without affecting the ingredient and the original form of the

of transmitted ray (Lim & Barigou, 2004). Thus, the transmission level

food. Many different noninvasive techniques have been studied and

of X-ray is determined by the mass as well as the absorption coefficient

developed for detecting foreign bodies. However, each technique has

of a sample. The X-ray technique consists of two types which are pla-

its own advantages and limitations under different application condi-

nar and computed tomography (CT). The planar type produced a two-

tions. There have been some reviews of this topic such as in Graves,

dimensional (2D) image after an X-ray generator emitted the ray to the

MOHD KHAIRI

|

ET AL.

3 of 20

object in a straight-line position. In the planar type, the X-ray absorp-

the food product. Several round shaped foreign bodies (stainless steel,

tion level is analyzed. On the contrary, CT used digital geometry proc-

Teflon, aluminum, rubber, glass, and ceramics) with different diameters

essing to generate a three-dimensional image of the inside of an object

were used. The minimum and maximum diameters are 0.3 mm and

from a large series of one or two-dimensional X-ray images taken

8 mm, respectively. The results showed that the detection rate for for-

around a single axis of rotation (Mathiassen et al., 2011). X-ray is

eign objects was above 98% without false positives. However, the

widely used for many applications such as construction, for example,

detection rate is low for small diameter (0.3–5 mm) as well as low den-

cement inspection and crack detection (Garboczi, 2002; Wang, Frost,

sity materials such as Teflon (2.18 g/cm3) and rubber (1.30 g/cm3).

Voyiadjis, & Harman, 2003), to detect tumors and bone crack for medi-

Mery et al. (2011) used X-ray machine vision technique to detect

cal application (Bandyopadhyay, Biswas, & Bhattacharya, 2016; Meinel

fish bones in fish fillets. The technique has several phases which begins

et al., 2014), painting analysis (Debastiani et al., 2016; Sun et al., 2014)

from image acquisition, followed by preprocessing, segmentation, fea-

as well as food industry, for example, quality control (Haff & Toyofuku,

ture extraction, classification, and finally post-processing. Salmon and

2008; Nicolaï et al., 2014) and characterization (Mousavi, Miri, Cox, &

trout fillet products were used as food samples. Twenty samples of

Fryer, 2005; Schoeman, Williams, Plessis, & Du Manley, 2016). X-ray

salmon fillets were tested in which the average size of these fillets was

has an edge in detecting food contamination as its ray has a stronger

15 3 10 cm2. The ranges for fish bones were categorized as large size

penetrating power. The advantages of using the X-ray method are that

(larger than 12 mm), medium size (between 8.5 and 12 mm), and small

it can inspect the food product based on nondestructive measurement

size (smaller than 8.5 mm). The results showed that the percentages of

and can offer a high imaging resolution. Conversely, X-ray has several

cross validation detection rate achieved were 100, 98.5, and 93.5% for

disadvantages such as high cost and high power usage (Haff &

large, medium, and small fish bones, respectively. The trout fillet was

Toyofuku, 2008; Pallav, Diamond, Hutchins, Green, & Gan, 2009).

tested using fish bones ranging from 14 to 47 mm and showed 99%

There is a negative perception that X-ray radiation give a bad effect to

percentage of detection.

the food. However, several studies stated that radiation levels from 7.5

A transmission and dark-field X-ray imaging with a grating interfer-

to 10 kGray used in food inspection are extremely low and do not

ometer was developed by Nielsen, Lauridsen, Christensen, and

affect food’s nutritional value and is safe to eat (Ashley et al., 2004;

Feidenhans’l (2013) to detect foreign bodies in food. The grating inter-

FAO/IAEA/WHO, 1999; Tauxe, 2001).

ferometer has an ability to record simultaneously the transmission,

In food safety inspection, detection of foreign bodies on diverse

phase-contrast, and dark-field images for obtaining the multiple inde-

foods has been investigated. Morita, Ogawa, Thai, and Tanaka (2003)

pendent mechanisms of contrast. Minced beef and a cultured sour

applied a soft X-ray generator to detect various types of foreign bodies

cream were used as food product which were mixed with three differ-

in a loaf of bread, a hamburger steak and cabbage. Six types of foreign

ent types of foreign materials. Glass, four layers of papers and a lady-

bodies namely steel screws, aluminum rivets, staples, aluminum foil,

bug were inserted into the minced beef. The cultured sour cream was

glass, plastic fragments, and grasshopper were used in this experiment.

mixed with eight layers of paper, a cigarette butt and a fly. Contrast-to-

The results showed that the metallic and nonmetallic types of foreign

noise ratio (CNR) measurement is used to analyze the contrast

materials contained in a load of bread and hamburger steak could be

between foreign bodies and the food product where a high CNR value

detected by the soft X-ray measurement system. However, it cannot

indicated that the contrast between them is significantly higher than

detect the plastic fragments. The grasshopper image under the cabbage

the noise in the image. In minced meat, the dark-field image shows

leaves was identified using various filter methods such as sobel,

higher CNR value for papers and ladybug. But for glass, a CNR value of

unsharped masking, and high pass filter.

2.3 was obtained. Conversely, the result shows that the dark-field

Foreign bodies detection in sealed chili soup packages was investi-

image gives a high CNR for all foreign bodies mixed with the sour

gated by Chen, Jing, Tao, and Cheng (2005) using a real time X-ray

cream than the transmission image. Papers, cigarette butt, and a fly

imaging method. The measurement system used a combination of edge

resulted in CNR values of 2.8, 6.3, and 2.9 for dark-field image com-

detection, region growing, and blob analysis techniques to obtain high

pared to the transmission image which gave CNR values of 0.4, 0.7,

quality images. Metals and bone fragments were mixed with chili soup

and 0.2, respectively.

bags and the scanning process was performed using an X-ray imaging

A dual-view X-ray inspection system to identify glass fragments in

system which consisted of an X-ray camera, an image processing board,

a glass jar was developed by Lu and Peng (2013). The scanning process

an X-ray tube, and a control unit. Three parameters in the canny edge

of the glass jar was performed by placing two X-ray generators in two

detector, that is, standard deviation (r), high and low thresholds were

orthogonal directions. The method was able to solve the blind area

varied to evaluate the strong and weak edges. The standard deviation

problem as it generated two pairs of X-ray images. Algorithms such as

was set between 1 and 2, while the high and low threshold values

adaptive image segmentation based on contour tracking and nonlinear

were set between 0.6 and 0.8 and 0.3 and 0.5, respectively. The combi-

arctan function transform were implemented to obtain a high detection

nation technique successfully distinguished the foreign bodies con-

sensitivity and high-quality images. Five glass fragments of different

tained in the chili soup.

sizes from 3 to 10 mm were placed at the top, middle and bottom of

Kwon, Lee, and Kim (2008) classified the foreign bodies in pack-

the jar. The results showed that the measurement system was capable

aged food using a real time X-ray imaging method. Three types of pack-

of generating a high-quality images of the jar and glass fragments and

aged dry foods; instant ramen, macaroni, and spaghetti were used as

the average detection rate was 92.9%.

4 of 20

|

MOHD KHAIRI

ET AL.

F I G U R E 1 (a) Image of the milk powder contaminated with the polyethylene (PE) plastic. (b) Image of the hollow cylinder inserted in minced pork meat. (c) Image of the hay powder mixed with milk powder (Li et al., 2015)

Niemeyer (2015) investigated food contaminated with foreign

mixed with the minced pork meat. The contaminated foods were

objects by utilizing the scanning electron microscopy (SEM) in conjunc-

visualized using confocal 3D Compton scatter tomography as shown in

tion with energy dispersive X-ray spectrometry (EDS). The EDS analysis

Figure 1a–c. The images show the foreign bodies which adulterated

provided a qualitative spectrum which revealed the food components.

the food product. Figure 1a shows the X-ray image of layers of the PE

Two cases were studied; bakery products adulterated with glass frag-

plastic inserted in the milk powder where the gray level reflected the

ment and powder seasoning mixed with metal particles. The EDS analy-

changes in the X-ray intensity and different densities of samples. Figure

sis demonstrated that the foreign materials can be discovered by

1b,c shows the X-ray images of the hollow cylinder inserted in the

showing the material components inside the food based on the qualita-

minced pork meat and hay powder adulterated in the milk powder,

tive spectrum.

respectively.

Li et al. (2015) applied a confocal micro X-ray scattering technol-

ttir et al. (2016) used grating-based multimodal X-ray Einarsdo

ogy based on polycapillary X-ray lens to detect foreign bodies in food.

imaging to visualize several types of foreign bodies. Seven food sam-

The technique used a polycapillary focusing X-ray lens (PFXRL) in the

ples, that is, minced meat, steak, turkey schnitzel, salami slices, sliced

excitation channel and a polycapillary parallel X-ray lens (PPXRL) in the

cheese, wheat bread, and rye bread were adulterated with eight differ-

detection channel. Two types of food samples were tested, that is, milk

ent foreign bodies (glass, metal, wood, insects, hard plastic, soft plastic,

powder and minced pork meat. The experiment was conducted using

rubber, and stones). This grating-based technique provided three imag-

foreign bodies which had a low density and low atomic number (Z) of

ing modalities; conventional absorption X-ray, phase contrast imaging,

chemical element. A polyethylene (PE) plastic, hay powder, and a hol-

and dark-field imaging which generate improved contrast capabilities

low cylinder were selected as foreign bodies. The PE plastic has a den-

compared to the conventional X-ray absorption imaging technique. The

sity of 0.926–0.940 g/cm3 and a dimension of 2 3 1 3 2 mm, whereas

Expectation-Maximization (EM) algorithm was developed to compare

the hollow cylinder has a density of 1.19 g/cm3, 15 mm external diame-

the X-ray imaging modality detection results and to determine the gain

ter, and 5 mm internal diameter. A PE plastic and hay powder were

of multivariate and texture analysis. Figure 2 shows the results when

used to contaminate the milk powder, whereas a hollow cylinder was

turkey schnitzel products were contaminated with eight foreign bodies.

MOHD KHAIRI

|

ET AL.

5 of 20

F I G U R E 2 Images obtained for turkey schnitzel products with foreign bodies. From left to right: absorption, phase contrast, and dark-field  ttir et al., 2016) imaging (Einarsdo

The level of contrasts between these three modalities were obtained

A thermal imaging system is classified into two types; active or

by normalizing gray profiles at the 250th row of each image. The dark-

passive thermal imaging systems. An active system requires the appli-

field model has the most contrast as it was able to display all foreign

cation of thermal energy to produce a thermal contrast between the

bodies.

features of interest and the background. In contrast, a passive system does not need to apply any external energy to the object because the

2.2 | Thermal imaging

features of interest are naturally at a higher or lower temperature than the background (Chen et al., 2013). The thermal imaging technique was

Thermal energy is the part of electromagnetic radiation within the

initially applied in military applications where it can be used to detect

range of 0.78–1,000 mm. In thermal imaging, the radiation pattern of an

the presence of human especially at night. Then, the application is used

object is transformed into visible images (Vadivambal & Jayas, 2011).

in other fields such as medicine (Kateb, Yamamoto, Yu, Grundfest, &

The temperature and emissivity of objects affected the level of

Gruen, 2009; Lahiri, Bagavathiappan, Jayakumar, & Philip, 2012), fire

released radiation (Orina, Manley, & Williams, 2017). Thermal imaging

safety (Amon, Hamins, Bryner, & Rowe, 2008), agriculture (Baranowski,

systems typically consists of a camera, an optical system (focusing lens,

Mazurek, Wozniak, & Majewska, 2012; Mangus, Sharda, & Zhang,

collimating lenses, and filters), a detector array, signal processing, and

2016; Stajnko, Lakota, & Hočevar, 2004), and food industry (Gowen

an image-processing system (Gowen, Tiwari, Cullen, McDonnell, &

et al., 2010; Jha et al., 2011). In the medical field, thermal imaging is

O’Donnell, 2010). An example of a measurement setup for the system

used to diagnose cancer, dental, blood pressure, and fever. Meanwhile

is shown in Figure 3.

in the fire safety field, thermal imaging has been successfully used to detect hot spots, search/rescue operation and identify the location of hazardous materials. In agriculture, it is widely used to monitor fruit maturity, predicting fruit yield, and detecting fruit disease. Thermal imaging is useful in various applications as it is a noncontact measurement, does not emit harmful radiation and can operate in real-time (Gowen et al., 2010). However, the technique needs to overcome temperature interference from other surfaces which has restricted its applications (Jha et al., 2011). In the food industries, Meinlschmidt and Maergner (2002) successfully developed an infrared thermography to detect foreign bodies in

An example of a thermal imaging measurement setup. (A) operating terminal; (B) thermal camera; (C) heat lamps; (D) sample stage; (E) USB power relay for lamp control (Kuzy & Li, 2017) FIGURE 3

raisins and almonds product. The thermography system consists of a heating and a cooling unit, a thermographic camera, an image processing unit, and a conveyor belt. Several wooden sticks and stones were used as foreign bodies and were mixed with the raisins and almonds

6 of 20

|

MOHD KHAIRI

ET AL.

product, respectively. They used the Thermosensorik-System CMT 384 M4 thermographic camera which can detect 3.4–5.2 mm of infrared radiation range. The emissivity coefficients and heat conductivity were analyzed to distinguish the foreign bodies. The heat conductivity technique provided better results than the emissivity coefficients as the value of emissivity coefficients between the food products and foreign materials are very close. The heat conductivity technique was applied by adding an extra heat source to the product when the product moved on the conveyor belt. The images of the food product and foreign bodies were obtained based on the gray levels value. Ginesu et al. (2004) continued the research using the same measurement system. They successfully improved the images by applying an interactive selection of the best sequence of image processing operations. The preprocessing started with a dead pixel correction, followed

An example of near-infrared spectroscopy measurement setup (Kamruzzaman et al., 2011)

FIGURE 4

by first enhancement filter application, second enhancement filter application, shading correction, and finally histogram stretching. The correction of dead pixels was performed by substituting each dead pixel value with a correct one. Convolution, rank, or morphological filters were implemented for the first enhancement filter, followed by a median filter for the second enhancement filter. Next, the shading problem was corrected by acquiring a background image and subtracted from the original one. Lastly, a histogram stretching was performed for rendering the image. The algorithm implementation based on local thresholding, rank order statistics, and morphological operators showed an improvement on the image quality index between the food product and foreign bodies. Bukowska-Belniak et al. (2010) evaluated a chocolate bar contaminated with several foreign bodies. Stone, plastic, and glass fragments were located at different depths below the surface of chocolate bar. The chocolate bar was heated up to a temperature level of 26 8C, and then, it was cooled down to 16 8C before the images were captured using a thermographic camera. Image processing was carried out as follows; first, object detection was based on images chosen via visual observation, and second, the results were used to identify the most

2.3 | Near-infrared spectroscopy and hyperspectral imaging Near Infrared spectroscopy (NIR) is a type of vibrational spectroscopy that produced photon energy in the range of 2.65 3 10219 to 7.96 3 10220 Joule which is the range of wavelength for near-infrared light from 750 to 2,500 nm (Wang & Paliwal, 2007). The bond vibrations between the atoms of organic molecules such as OAH, CAH, and NAH caused the change of energy when exposed to NIR light (Li, Sun, & Cheng, 2016). The resulting patterns of absorption/reflection across the wavelength can be used to obtain the characteristic and features of the material tested. NIR spectral imaging has been successfully applied to several food types such as fish (Mathiassen et al., 2011), milk (Huang et al., 2016), rice (Kong, Zhang, Liu, Nie, & He, 2013), and lamb (Kamruzzaman, Elmasry, Sun, & Allen, 2011) for analyzing their quality and texture. Figure 4 shows an example of the NIR measurement setup. It consists of a camera for capturing the image, a spectrograph to separate light into a frequency spectrum, lens for adjusting the receiving light, an illumination unit for determining the spectral range of the sys-

contrastive images. The results demonstrated that the contaminants’

tem, a translation stage for putting samples of food and a motor for

materials can be detected successfully by generating a sequence of

moving the food (Kamruzzaman et al., 2011). There are two categories

thermographic images.

of light sources for spectrometric measurements; thermal sources and

Research on the detection of foreign bodies in biscuits using infra-

nonthermal source (Osborne, Fearn, & Hindle, 1993). Thermal sources

red thermography was carried out by Senni et al. (2014). A comparative

such as tungsten halogen lamps produce radiation spanning a continu-

analysis of the thermal emissivity of foreign bodies and of biscuits

ous spectral region. However, this type of source disturbs the tempera-

decay was performed. Biscuits’ dough was contaminated with several

ture stability of the testing material (Butz, Hofmann, & Tauscher, 2006).

types of foreign bodies such as stone, glass, plastic, wood, paper, and

Light emitting diodes (LED) and lasers are some examples of nonthermal

textile fiber. The size range of the foreign bodies was from 1 to 3 mm.

sources where the sources produce radiation within a narrow spectral

The dough was cooked for 6 min at a temperature around 250 8C. Hun-

band. It is low cost, portable, and has variation of emission wavelengths

dred and seventy thermal images was collected after the biscuits

(McClure, Moody, Stanfield, & Kinoshita, 2002).

cooled down and reached the room temperature. The result demon-

NIR has several advantages such as it is a nonionizing technique

strated that most of the foreign bodies can be detected by the ther-

and its ability to penetrate air gaps within the food materials (Pallav,

mography measurement system. They also used a thermocamera

Diamond, et al., 2009). NIR is also capable of discovering a small ele-

model FLIR A-315 to perform on-line measurement where it has a

ment in the food internal structure as it can capture the image in the

maximum frame rate of 60 Hz and it was able to cover an area of

wavelength’s nanometer range (Tsuta, Takao, Sugiyama, Wada, &

25 cm 3 20 cm. The generated images were normalized before a two-

Sagara, 2006). NIR computes a mean spectrum of a sample and pro-

fold thresholding procedure was applied. Finally, the biscuits and for-

vides a single spectrum, but the information data may be insufficient

eign bodies were classified.

and complex to analyze. NIR has some limitations as it depends on

MOHD KHAIRI

|

ET AL.

7 of 20

F I G U R E 5 (a) A block cheese mixed with a 10 mm rubber piece. (b) Meat sample containing an absorbent Blu-Tack material (Pallav, Diamond, et al., 2009)

reference methods for calibration purpose (Jha et al., 2011; Manley,

system used 3 kHz sinusoidal modulation and has 850 nm of wavelength

2014). A technique called hyperspectral is proposed which made use of

which was able to detect a nonmetallic material. The images were gener-

a mixture of red green blue (RGB) and grayscale (Huang, Liu, & Ngadi,

ated based on data from the transmission mode technique. Several types

2014). The hyperspectral image (HSI) produced three axes where the x-

of food samples such as dough, cheese block, meat, and chocolate bar

axis and the y-axis represent the spatial dimension, while the z-axis

were contaminated with different types of foreign bodies (coin, glass

denotes the wavelength (Liu, Pu, & Sun, 2017). The hyperspectral anal-

ball, and rubber). Figure 5a shows the image produced when the block

ysis involved the acquisition of a heap image of the same object at dif-

cheese was mixed with a 10 mm rubber piece. Figure 5b illustrates a

ferent spectral bands. In HSI, samples experienced different amount of

meat sample containing an absorbent Blu-Tack material. The meat was

scattering, reflection, absorption, and emission of electromagnetic

placed in a plastic packaging and consists of a layer of paper at the base

energy

structures

of the container. It was observed that at a wavelength of 850 nm, NIR

(Moghaddam, Razavi, & Taghizadeh, 2013; Siche et al., 2016). The ana-

signals were easily transmitted across the meat sample. The highly

lytical information from the spectra is extracted and analyzed through

attenuating Blu-Tack was detected with a high contrast which is visible

the application of multivariate analysis tools such as artificial neutral

as the white area in the NIR image of Figure 5b. The results revealed

networks and principal component analysis to expose the differences

that the NIR system was able to generate the image which is very useful

between analyzed samples (Orina et al., 2017). HSI has the advantages

in monitoring food condition and food quality.

depending

on

the

physical

and

chemical

of being able to provide spatial and spectral information. It is also sensi-

NIR spectral imaging was applied by Sugiyama et al. (2010) to

tive to minor components. However, the limitation of HSI is the

observe leaves and stems in blueberry product. Discriminant analysis of

lengthy time required for pre-processing of data and classification

absorbance spectra was used to determine the optimal illumination

(Senni, Burrascano, & Ricci, 2016; Xiong, Sun, Zeng, & Xie, 2014).

wavelengths of foreign materials. Two values, that is, 1,268 nanometer

Tsuta et al. (2006) developed a foreign materials detection method

(nm) and 1,317 nm were obtained. The absorbance images of blueber-

for blueberries product using a spectral imaging technique. Foreign

ries contained in the leaves and stems were generated based on these

materials such as leaves, twigs, and stones were dyed in the same color

values. The blueberry samples were stored at 218 8C to maintain their

as blueberries by soaking them in the blueberry juice. Spectral images

quality and the samples for investigation were selected based on their

were acquired at 660, 680, and 700 nanometer (nm) to develop a sec-

maturity and size. The images were produced from five samples for

ond derivative absorbance image at 680 nm. Fifty two blueberries sam-

each of the three sample types (blueberry, leaf and blueberry, stem,

ples and 26 foreign materials were used for the analysis where the

and blueberry). The canonical discriminant function and threshold value

second derivative absorbance and mean values were calculated. The

for image binarization were obtained by applying the discriminant anal-

probability of a pixel containing a foreign material was measured and a

ysis on the absorbance images. The results in Figure 6 shows binarized

pixel larger than 0.95 was considered as a foreign material. The results

images of the leaves and stems which were mixed with the blueberries.

showed that the foreign materials can be detected in the blueberry

The blueberry surface, the leaf, and the stem are clearly distinguished

sample.

from each other in the binary image. All canonical images using absorb-

A NIR imaging system was developed by Pallav, Diamond, et al.

ance images were taken at 1,268 and 1,317 nm. The threshold value

(2009) to detect foreign bodies in several types of foods. The NIR

for binarization is 3.949. In the linear discriminant analysis, the

8 of 20

|

FIGURE 6

MOHD KHAIRI

ET AL.

Image binarization of blueberries with leaf and stem (Sugiyama et al., 2010)

discriminant function is calculated using spectral information on each

the same NIR measurement system and successfully detected a

pixel only, and pixel locations are not used for distinction. Thus, the

wooden stick having a diameter of 0.3 mm in ham slices.

accuracy of the discriminant analysis is not affected by the disposition

Jiang, Zhu, Rao, Berney, and Tao (2007) used hyperspectral fluo-

or number of foreign materials in the same image. The results show

rescence imaging to visualize shell and pulp in black walnuts product.

that the accuracy of the discriminant function is sufficient for distin-

The walnuts images obtained from the hyperspectral system were clas-

guishing the leaves and the stems from the blueberries.

sified using three different techniques; Gaussian-kernel based support

Foreign bodies contained in ham slices and chocolate were suc-

vector machine (SVM), principal component analysis (PCA), and Fisher’s

cessfully distinguished by a NIR imaging system developed by Tashima

discriminant analysis (FDA). Four samples were categorized based on

et al. (2013). The NIR system has 40 superluminescent diodes (SLDs) as

four conditions, that is, light pulp, dark pulp, inside shell, and outside

light sources and a wide dynamic range CMOS sensor. The ham slices

shell. It was easy to visualize light pulp and inside shell. Conversely, it is

and chocolates were contaminated by inserting hairs (0.1 mm in diame-

difficult to visualize dark pulp and outside shell. Six thousand two hun-

ter) and insects (3 mm in width) at a depth of 2.6 mm from the surfa-

dred and fifty seven data sets were used for training and testing sam-

ces. The image contrasts between the foods and foreign bodies were

ples in the statistical analysis. The SVM technique was found to be the

improved and sharpened by cutting out the low spatial frequency data.

best approach in classifying the shell and pulp of walnut with a recogni-

The foreign substances were clearly distinguished from the foods in

tion rate of 90.3% compared to PCA and FDA which have recognition

the images. Phetchalern et al. (2014) continued the investigation using

rates of 87.7 and 85.8%, respectively.

MOHD KHAIRI

|

ET AL.

9 of 20

R2 5 .992. The value is higher compared to the automatic speck counter which has R2 5 .767.  Díaz, Cervera, Fenollosa, Avila, and Belenguer (2011) used a hyperspectral system for detecting foreign materials in pork steaks products. The hyperspectral system consisted of an infrared camera, a spectrograph and four infrared halogen lamps. Several types of foreign bodies such as polyethylene terephthalate (PET), polyethylene (PE), metal, insects, and bone were prepared with different compositions and sizes of 2 3 2 mm, 5 3 5 mm, and 10 3 10 mm. The images were obtained, and a multivariate analysis was applied to evaluate the data sets and images of each sample. The principle components analysis (PCA) was applied to select the best data reduction and discrimination abilities. The result obtained by PCA was used in the Fischer discriminant analysis to classify each point of the sample. A new image was generated after implementing the discriminant functions and the results showed that most of the evaluated materials can be detected. A NIR hyperspectral imaging system also was applied by Gowen & F I G U R E 7 Correlation graph between the seeded insect fragments level and the predicted insect fragment level (Bhuvaneswari et al., 2011)

O’Donnell (2013) to identify and classify foreign bodies in grains product. The hyperspectral imaging system was operated in the diffuse reflection mode in the wavelength range of 950–1,700 nm. Two types of grain samples were assessed which is a uniform sample of white rice

Bones’ fragments contained in chicken breast fillets were investi-

grains and a mixed variety grain sample. Plastic shards, glass beads and

gated by Yoon, Lawrence, Smith, Park, and Windham (2008) using a

rubber fragments were used as foreign bodies. The glass samples

hyperspectral imaging (HSI) technique. The HSI method which made

exhibited a very low spectral response whereas the plastic and rubber

use of a combination of transmission and reflection was used and the

samples exhibited characteristic absorbance of near infrared (NIR) radi-

contrast of the images is very low due to the light scattering effects.

ation at certain wavelengths. The maximal variance projection was

The images were improved by applying an illumination-transmittance

obtained using the principal components analysis (PCA) where the data

model to simplify bones detection using a single threshold. The image

was used for detection and identification of foreign bodies. The study

processing algorithms incorporated histogram stretching, thresholding,

indicated that the NIR reflectance approach could detect and identify

median filtering, and image fusion. The hyperspectral camera system

all types of foreign bodies among grain samples.

that was used in this investigation consisted of a spectrograph, a 12-bit CCD camera, a focal plane scanner, and a front lens. The average size

2.4 | Ultrasonic

of the bones’ fragment was about 2.4-cm long and 0.2-cm thick and they were inserted in twenty meat samples. The results showed that

Ultrasound is referred to as sound waves above the frequencies of

the measurement system has a false-positive rate of 10%.

audible sound, nominally greater than 20 kHz. (Awad, Moharram, Shalt-

Bhuvaneswari et al. (2011) compared the NIR hyperspectral imag-

out, Asker, & Youssef, 2012). An ultrasonic wave is generated when

ing technique with the automatic speck counter technique to evaluate

the elastic deformation in ferroelectric material is in high frequency

semolina product contaminated with insect fragments. In their experi-

range. A high-frequency current was transmitted via two electrodes to

ment, they analyzed the insects based on the specks results obtained

the ferroelectrical material which subsequently produced an ultrasonic

from both techniques. The fragments of Tribolium castaneum species

wave (Knorr, Zenker, Heinz, & Lee, 2004). Ultrasonic sensing is one of

were prepared at 0, 50, 75, 150, and 300 fragments per 50 g of semo-

the techniques that attracted interest in the food industry and has

lina. The NIR hyperspectral imaging system was implemented by scan-

been successfully used in several applications such as cavitation

ning an area of 200 3 200 pixels and the distributed wavelengths is in

(Bhaskaracharya, Kentish, & Ashokkumar, 2009; Chandrapala, Oliver,

the range of 1,000–1,600 nm. The mean reflectance spectrum was

Kentish, & Ashokkumar, 2012a; Kentish & Feng, 2014; Shanmugam,

measured by averaging the reflectance intensity values at each of the

Chandrapala, & Ashokkumar, 2012), emulsification (Abbas, Hayat,

61 wavelengths. Partial least-square (PLS) was integrated with the NIR

Karangwa, Bashari, & Zhang, 2013; Freitas, Hielscher, Merkle, &

hyperspectral imaging to evaluate semolina product contaminated with

Gander, 2006; Leong, Wooster, Kentish, & Ashokkumar, 2009;

insect fragments. The use of partial least-square (PLS) regression

Mongenot, Charrier, & Chalier, 2000), and freezing (Chemat, Zill-

showed that a good correlation was obtained between the insect frag-

e-Huma, & Khan, 2011; Sigfusson, Ziegler, & Coupland, 2004; Zheng &

ments in the semolina and the specks value as shown in Figure 7.

Sun, 2006). The sensing modes for the ultrasonic sensor are the trans-

When there were no insect fragments, the graph slope has a value of

mission, reflection and diffraction modes. The receiver is located in

0.893. Then, the test was performed by increasing the number of frag-

front of the transmitter for the transmission mode. For the reflection

ments in semolina and a good response was obtained in which

mode, the ultrasonic transmitter and receiver are located at the same

10 of 20

|

FIGURE 8

MOHD KHAIRI

ET AL.

€m & Luukkala, 2001) The ultrasonic measurement setup (Hæeggstro

side. For the diffraction mode, the receiver should be placed in a suita-

Zhao, Basir, and Mittal (2003) investigated various sizes of glass,

ble position to obtain a precise diffracted signal. The change in wave-

metal and plastic in bottled beverages using an ultrasonic sensor with a

form detected by the receiver can provide valuable information such as

nominal frequency of 15 MHz. The ultrasonic transducer was placed at

attenuation, changes of velocity, time of flight, and acoustic impedan-

the bottom of the container and the wave was transmitted in an

ces (Alava et al., 2007; Chandrapala, Oliver, Kentish, & Ashokkumar,

upward direction. The tested frequencies were in the range from 5 to

2012b). These parameters are exploited to investigate and analyze the

10 MHz. A signal processing algorithm for time–frequency analysis was

food properties. There are several advantages in using ultrasound in

used to discriminate between the beverage and the foreign substances.

the production line as it is a nondestructive sensing technique, it has a

Two ultrasonic echoes were evaluated to determine the size of foreign

fast response and it is able to perform high-velocity measurement

bodies; the reflection waves at the outer container surface and the

(Hauptmann, Hoppe, & Puttmer, 2002; Henning & Rautenberg, 2006;

reflection wave at the inner surface of the container. The pressure ratio

Mcclements & Sundaram, 1997). However, an ultrasonic sensor is

between the two echoes was analyzed. The pressure ratio was the low-

limited to a single point measurement and it is very sensitive to the air

est value when there were no foreign bodies inside the container for

medium as the ultrasonic wave cannot propagate properly in the air.

frequencies up to 8 MHz. They also attempted to detect a glass frag-

This is due to a massive different of acoustic impedance between air

ment using an ultrasonic sensor based on a combination of the radial

and other media (Jha et al., 2011). Ultrasonic sensors have widely been

basis function neural network algorithm and the short-time Fourier

applied for determining and evaluating the quality and safety of food

transform (Zhao, Yang, Basir, & Mittal, 2006). The input features con-

such as investigations conducted by Chandrapala et al. (2012b),

sisted of parameters related to the glass fragment such as size, position,

Chandrapala and Leong (2015), Kim, Lee, Kim, and Cho (2009), Elvira

orientation and the incident angle of the ultrasonic beam. The number

et al. (2005) as well as Morrison and Abeyratne (2014).

of neurons were varied from one to seven and they obtained a classifi-

Nonimaging and imaging techniques have been applied for detect-

cation rate of 95% at seven neurons.

ing foreign bodies in food using ultrasonic sensor. For the nonimaging

Correia, Mittal, and Basir (2008) applied an ultrasonic sensing sys-

technique, the evaluation and analysis is based on ultrasonic parame-

tem to detect bone fragments in deboned chicken breast. The ultra-

ters such as sensing modes, velocity and attenuation. An ultrasonic

sonic transducer emitted a pulse of 4 MHz peak frequency. They used

measurement system based on the reflection sensing mode was devel-

a piston and cylinder apparatus to perform pulse-echo based ultrasonic

€m and Luukkala (2001) to investigate the presence oped by Hæaeggstro

measurements for solid samples of variable heights. The bone frag-

of foreign bodies. The measurement setup as shown in Figure 8 con-

ments sample was categorized as large (15.75 mm2), medium

sisted of a 5 MHz ultrasonic transducer, an oscilloscope, a preamplifier,

(9.92 mm2), and small (6.18 mm2). Several parameters such as velocity,

a computer, a pulser, and a receiver. They successfully detected various

impedance and amplitude ratio were used to determine the size of

sizes of bones, glass, steel as well as wood in marmalade and cheese

bone fragment. From these parameters, the amplitude ratio showed a

product. The food product was evaluated within the temperature range

better accuracy compared to other parameters. The system success-

of 0.5 to 22 8C where ice was used to control the temperature in the

fully classified the uncut samples, cut samples, and cut samples of the

basin. The results showed that all foreign bodies can be detected in all

bone fragments with a projected area of 6 to 16 mm2.

kinds of food products. However, they have difficulties in determining

Leemans and Destain (2009) evaluated foreign bodies contained in

the types of foreign bodies whether it is bone, steel or glass. The best

a semi-soft cheese based on the time of flight of the transmitted signals

result was obtained when the Naturelle cheese brand was assessed

and the echo signals. A plastic cylinder with a diameter of 3 mm was

where all types of foreign bodies could be detected and identified.

inserted into the cheese. The experiment was conducted in varying

MOHD KHAIRI

ET AL.

|

11 of 20

F I G U R E 9 The tomographic images for the cans inserted by (a) a copper rod with a diameter of 6.6 mm (b) two aluminum rods each has a diameter of 7.2 mm (Ho et al., 2007)

temperatures from 4 to 17 8C to observe the accuracy of the measure-

receiver sensors by adding the EMAT coil to generate and detect the

ment process. Signal-noise-ratio (SNR) values at different temperatures

ultrasonic wave’s surrounding the cans. The pulse compression tech-

for transmission, echo of the opposite face of the cheese and echo due

nique was used to improve the signal-noise-ratio (SNR) and images

to the foreign body were evaluated and the results showed that the

inside the cans were reconstructed using the tomography method. The

system has a recognition rate of 90%.

measurement system was tested using a copper rod which has a diam-

Meftah and Mohd Azimin (2012) used pulse-echo ultrasonic test-

eter of 6.6 mm and two aluminum rods in which each has a diameter

ing to detect foreign bodies in canned food. The evaluation was per-

of 7.2 mm. Both materials were used as foreign bodies. The fan beam

formed using three types of foreign bodies having different sizes; a

technique was applied to reconstruct the image inside the can. The fan

rock (62 mm 3 38 mm) and two aluminum plates (30 mm 3 23 mm 3

beam technique demonstrated that several receivers had detected the

4 mm and 122 mm 3 21 mm 3 1 mm) were placed inside an aluminum

beam from a transmitter. Figure 9a shows the image of the cylindrical

tin container filled with water. An ultrasonic transducer of 4 MHz

container filled with water with a 6.6 mm diameter copper rod. The rod

placed on the container wall transmitted the wave and the wave was

is located at the coordinate (55, 35) mm. The images of the copper rod

reflected when it collided with the foreign bodies. The results showed

can be observed as the wave could not propagate directly through the

that the measurement system can detect the foreign bodies with a size

rod. The experiment was conducted with two 7.2 mm diameter alumin-

range of 1–4 mm.

ium rods located separately at the coordinates (15,33) mm and (48,48)

A noncontact ultrasonic imaging technique was proposed by Cho and Irudayaraj (2003) to distinguish foreign bodies in cheese and poul-

mm. The reconstructed image is shown in Figure 9b which shows the position of the aluminium rod accurately despite the slight distortion.

try product. The measurement system used a pair of 1 MHz ultrasound

An ultrasonic imaging system have been developed by Pallav, Hutchins,

transducers, where the velocity and the attenuation of the ultrasound

and Gan (2009) to evaluate the presence of foreign bodies in cheese

waves were enhanced using a noncontact air instability compensation

products. The noncontact ultrasonic transmission mode technique was

method. The method was performed by installing a ring shape refer-

implemented. A pulse compression technique was developed to obtain

ence in front of both transducer. It was used to monitor the air prop-

an improved SNR ratio and measurement timing for solving the coupling

erty changes of the air column between a transducer and a reference

issue in air-couple measurement. Wood, rubber, and glass were used as

in real time. The air temperature was varied from 26.0 to 28.5 8C to

foreign materials and each type was placed separately in the cheese prod-

analyze the measuring system performance. Various sizes of metal rod,

uct to investigate the reliability of the measurement system. The detec-

metal fragments and a glass fragment were inserted to the cheese and

tion capability of the ultrasonic images was enhanced by forming images

poultry product. Image of the food product were constructed and the

from a combination of both amplitude and time-of-flight data. It was

results showed that the metal and glass fragments which have a size of

required to overcome the lack of discrimination between the foreign

3 mm 3 3 mm can be detected. The measurement system also suc-

bodies and the natural background variations in the cheese. The ampli-

cessfully identified a steel rod which has a length of 1.5 mm.

tude and time of flight were measured, and the cheese images were con-

Foreign bodies in aluminum beverage cans were successfully

structed based on these parameters to visualize the foreign substances.

detected by Ho, Billson, and Hutchins (2007) using electromagnetic

The scanning images as shown in Figure 10 shows that the measurement

acoustic transducers (EMAT). They modified the transmitter and

system is effective in detecting the foreign bodies in the cheese product.

12 of 20

|

MOHD KHAIRI

ET AL.

Images of a cheese sample containing of (a) rubber, (b) wood, and (c) glass, obtained using the ratio of the amplitude and timeof-flight data (Pallav, Hutchins, & Gan, 2009)

FIGURE 10

2.5 | Terahertz

values of refractive indices were obtained in the case of glass and stone which have values of 2.6 and 1.9, respectively.

Terahertz waves lie between the microwave and infrared regions of the electromagnetic spectrum in the frequency ranging between 0.3 and 10 THz and a wavelength range from 100 mm to 1 mm (Qin, Ying, & Xie, 2013). It can be operated either in the transmission or the reflection mode and it can be used in the time and spatial domain. The radiation energy ranging from 1 to 10 meV caused the molecules to vibrate and rotate which bring out the wavelength properties of different materials (Butz et al., 2006; Gowen, O’Sullivan, & O’Donnell, 2012). Terahertz waves have several advantages to be used as a nondestruc-

Kim et al. (2012) utilized a horn antenna in a continuous wave terahertz imaging system to improve the quality of spatial resolution. The horn antenna waveguide is smaller than the wavelength and it is able to provide a high transmission power. The measurement system has a frequency of 0.2 THz and its output power was larger than 10 mW. Several sizes of foreign bodies such as aluminum foil, metallic cubes, cubic stones, grasshopper and mealworms were placed in a flour sample. The application of a horn antenna demonstrated that it can detect the size of foreign bodies less than 1 mm compared to 4 mm without

tive test method as it generates low photon energy and it is able to

the horn antenna. Phantoms with shapes such as a star mark, a K logo,

penetrate various materials (Gowen et al., 2012; Guillet et al., 2014). In

squares, circles, and bars were created using aluminum foil, as shown in

addition, terahertz waves also have low spatial resolution and low

Figure 11a to observe the improvement of spatial resolution in the

signal-to-noise ratio (Kim et al., 2012). However, terahertz waves are

presence of the horn antenna. All shapes were attached to a sheet of

not appropriate for use in water medium as the signal is highly attenu-

paper to prevent movement in the flour. The thickness of the flour was

ated (Pallav, Diamond, et al., 2009). Terahertz waves are widely used in

4 mm and the diameter was 50 mm. The flour was fixed by plastic films

food quality monitoring such as moisture content determination (Chua

on both sides and the sample was located at the rear of the horn

et al., 2005; Parasoglou et al., 2009), prediction of sugar and alcoholic

antenna. Figure 11b,c shows the different qualities of the images con-

content (Jepsen, Møller, & Merbold, 2007), oil characterization

structed with a horn antenna and without a horn antenna, respectively

(Gorenflo et al., 2006; Jiusheng, 2010), and residue detection

when several shapes and sizes of aluminum foil were inserted in the

(Redo-Sanchez et al., 2011; Yuefang & Hongjian, 2010).

flour. The minimum detectable size without the horn antenna was

€ rdens and Koch (2008) used a pulsed teraIn food safety area, Jo

above 4 mm. Conversely, the phantoms with a size of less than 1 mm

hertz imaging system to detect the presence of foreign bodies in choc-

can be detected using the horn antenna. Thus, this method can suffi-

olate. The experiment was conducted using a single pulse structure and

ciently detect objects in thicker samples. The other types of foreign

a double pulse structure. A single pulse structure was used to detect a

bodies (metallic cubes, cubic stones, grasshoppers and mealworms)

hazelnut as some chocolate product contains this ingredient. The dou-

were adulterated in the flour with a diameter of 50 mm as shown in

ble pulse structure was utilized to detect nonmetallic foreign bodies

Figure 12a,b shown the images obtained by the application of a horn

such as stone, glass and plastic fragments. All particles were buried

antenna.

inside the chocolate bar. The experiment was performed using the inte-

The comparison between the continuous wave (CW) terahertz and

grated intensities between 0.4 and 0.75 THz and the presence of for-

X-ray imaging techniques to visualize the foreign bodies in noodle was

eign bodies was evaluated based on the refractive index. The results

performed by Lee, Choi, Han, Woo, and Chun (2012). Several sizes of

showed that chocolate and hazelnuts have refractive indices of 1.75,

high and low-density materials were inserted into a powdered instant

while plastic fragments have a lower value which is 1.5. However, high

noodle to contaminate the product. For high density materials, an

MOHD KHAIRI

ET AL.

|

13 of 20

(a) Foreign bodies with several sizes and shapes of aluminum foils. (b) An image obtained by the application of a horn antenna. (c) An image obtained without the application of a horn antenna (Kim et al., 2012)

FIGURE 11

aluminum and granite had been selected, whereas for low density

placed inside the cracker. The images for the cracker were captured

materials, insects such as maggots and crickets were chosen. The

using a pyroelectric array camera which has a video rate of 48 frames

results demonstrated that the terahertz technique was capable of

per second. The foreign bodies were successfully visualized by the

detecting all sizes of high density and low density foreign bodies. In

measurement system and the results showed that a high power tera-

contrast, the X-ray imaging technique cannot visualize a small size

hertz radiation can improve the quality of the reconstructed images.

maggot.

Ok, Choi, Park, and Chun (2012) proposed a sub-terahertz quasi-

A sub-terahertz electron cyclotron resonance maser (gyrotron) was

Bessel beam (QBB) to distinguish the existence of foreign bodies in

developed by Han, Park, Ahn, Lee, and Chun (2012) to visualize foreign

instant noodles. The beam profile of QBB was analyzed using the

bodies in a cracker product. The gyrotron is selected as it can generate

finite-difference time-domain (FDTD) and was compared to the knife

a high output power with good spatial patterns in a sub-THz region. It

edge method and the point scanning method. The measurement sys-

was operated in the cavity mode which is 0.2 THz having a 9 kV beam

tem consisted of a 210-GHz transmitter with an output power of 75

and 7.3 Tesla of the axial magnetic field strength. Two foreign bodies

mW, a Schottky diode which acted as a receiver and a conical horn

which are soft type (dried fish) and hard type (metal fragment) were

antenna. Two sizes of crickets with dimensions of 35 mm 3 5.5 mm

(a) Foreign bodies of different sizes and materials in a flour sample (i) metallic cubes (ii) cubic stones (iii) grasshoppers (iv) mealworms. (b) Images obtained by the application of a horn antenna (Kim et al., 2012)

FIGURE 12

14 of 20

|

MOHD KHAIRI

Image comparison between the sub-terahertz and the Xray imaging system showing (a) metals (b) insects (c) plastic (Ok et al., 2014)

TA BL E 1

Foreign body

Sub-terahertz

X-ray

ET AL.

insects and plastic. However, the contrast of the insects and plastic was very well been observed using the sub-terahertz transmission image. Lee and Lee (2014) utilized a terahertz imaging system to detect a metal razor and a rubber fragment in powdered milk. The sizes of the

(a) Metal

foreign bodies were between 2 and 10 mm. The source output was shaped into a line beam and an array detector was used to achieve the fast imaging process. Both types of foreign bodies could be detected at a resolution of 0.8 and 1 mm. However, the metal has a lower transmitted intensity compared to the rubber material due to higher absorption and scattering losses. Nonmetallic foreign bodies in packets of powder was examined by

(b) Insects

Ikari, Takahashi, Fukasawa, and Duling (2014) using a terahertz measurement system. The system consisted of a fiber coupled THz pulse generation and detection system, a high-speed THz beam scanner, and a data analysis algorithm. The maximum beam scan rate was 17 sweeps/s and the image with 100 3 100 pixels can be acquired in 10 s. A spatial correlation analysis of the waveform or a vector analysis (c) Plastic

was applied to enable automatic detection of the foreign bodies. Fifteen types of resins consisting of Teflon and Polyvinyl chloride (PVC) material were evaluated and the result showed that the terahertz measurement system can visualize all resins except the Teflon fragment. It is due to the terahertz characteristics of Teflon which are almost similar to that of the lactose powder. Yu et al. (2015) developed a high speed terahertz imaging system using a continuous wave of 0.3 THz to evaluate the presence of a caterpillar inside a chocolate product. They utilized an orthogonally polar-

and 50 mm 3 7 mm were mixed with the noodle floor. The knife edge

ized THz wave for real-time imaging purpose. The scanning process

method was initially investigated and the minimum spot size of QBB

was performed when the chocolate was placed in a moving conveyor

and Gaussian beam was obtained. The point scanning method was

belt which has a speed rate of 72 meters per minute. The imaging sys-

introduced to improve the image quality as the QBB cannot identify

tem successfully showed that the caterpillar can be detected even

the interference in the central core. The results showed that the QBB

though the chocolate was packaged using a paper case. The applica-

produced sharper images compared to the images formed using the

tions of foreign bodies detection in various food products are summar-

Gaussian beam.

ized in Table 2. The advantages and disadvantages of each technique

The research was extended by Ok, Kim, Chun, and Choi (2014) to

are tabulated in Table 3.

evaluate various sizes of low density (insect and polymer) and high density (metals) materials in a powdered milk. They developed a high-

3 | CHALLENGES AND FUTURE TRENDS

resolution raster sub-terahertz scan imaging system where a Gaussian beam focusing method was implemented to obtain the diffraction-

Recent applications of noninvasive techniques for food safety and

limited imaging. Images of foreign bodies were produced in the trans-

quality evaluation especially for detecting the foreign body are sum-

mission and the reflection mode. The images based on the transmission

marized in Table 2. However, the applications of the techniques are

mode demonstrated better results in classifying foreign bodies which

based on their advantages and disadvantages as presented in Table 3.

have low densities. A commercial X-ray inspection system (Intellisense

The limitations which exist in each technique should be overcome to

XIS-1300S) was used as a comparison with the sub-terahertz imaging

enhance their capabilities and widen their applications. Most of the

system. The X-ray system was operated at 60 kV and 3 mA with a

discussed techniques are related to image processing which has some

detection area of 256 3 256 mm , a pixel size of 400 mm and an inten-

constraint in real-time application as it has to deal with a large set of

sity depth of 4,096 (12 bits) gray value. Both sub-terahertz and X-ray

data. An advance image processing algorithm with an adequate merg-

systems can visualize the metal foreign bodies, but the sub-terahertz

ing process onto the hardware is needed to minimize the processing

image resolution is shown to be inferior to that of the X-ray image as

time. Problem also arises when the size/shape of the foreign body is

shown in Table 1(a). The results obtained in Table 1(b) and Table 1(c)

similar to the food materials because some reconstructed images can-

revealed that images produced by the X-ray transmission mode has a

not distinguish them. Hence, the use of artificial intelligent system

poor contrast and cannot distinguish the low-density materials, that is,

(e.g., neural network and neuro-fuzzy) could be utilized to classify the

2

MOHD KHAIRI

TA BL E 2

|

ET AL.

15 of 20

Applications of noninvasive techniques for detecting foreign bodies

Mode

Food products

Foreign bodies

References

X-ray

Loaf of bread, a hamburger steak, nd cabbage

Steel screws, aluminum rivets, staples, aluminum foil, glass and plastic fragments Metals and bone fragments Stainless steel, Teflon, aluminum, rubber, glass, and ceramics Bones Glass, paper, a ladybug, a cigarette butt, and a fly Glass fragments Glass fragment and metal particles

(Morita et al., 2003)

Chili soup Instant ramen, macaroni, and spaghetti Fish fillets Minced meat, cultured sour cream product Food jar Bakery product and powder seasoning Milk powder, minced meat Cheese, minced milk, wheat bread Thermal Imaging

Raisins, nuts, almonds Raisins, almonds, nuts Chocolate bar

(Chen et al., 2005) (Kwon et al., 2008) (Mery et al., 2011) (Nielsen et al., 2013) (Lu & Peng, 2013) (Niemeyer, 2015)

Polyethylene plastic, hay powder, and hollow cylinder Glass, metal, wood, insects, plastic, rubber, and stones

(Li et al., 2015)

Wooden sticks and stones Wooden sticks, stone, metal, and cardboard Stone, plastic, and glass fragments

(Meinlschmidt & Maergner, 2002) (Ginesu, Giusto, Märgner, & Meinlschmidt, 2004) (Bukowska-Belniak, Lesniak, Kiełkowski, & Michalski, 2010) (Senni et al., 2014)

 ttir et al., 2016) (Einarsdo

Biscuits

Stone, glass, plastic, wood, paper, and textile fiber

Blueberries Dough, cheese, doughnut, meat Blueberries Ham slice and chocolate Ham slice, fish, and chicken wing sticks Shell and pulp Chicken breast fillets Semolina Pork steaks

Leaves, twigs, and stones Coin, glass ball, and rubber Leaves and stems Hairs and insects Wooden sticks and bones

(Tsuta et al., 2006) (Pallav, Diamond, et al., 2009) (Sugiyama et al., 2010) (Tashima et al., 2013) (Phetchalern et al., 2014)

Walnut Bone fragments Insect fragments Polyethylene terephthalate, polyethylene, metal, insects, and bone

(Jiang et al., 2007) (Yoon et al., 2008) (Bhuvaneswari et al., 2011)  (Díaz, Cervera, Fenollosa, Avila, & Belenguer, 2011)

Hyperspectral imaging (HSI)

Grains

Plastic shards, glass beads, and rubber fragments

(Gowen & O’Donnell, 2013)

Ultrasonic

Marmalade and cheese product Bottled beverages Bottled beverages Deboned chicken Cheese Canned food Cheese and poultry product Canned beverages Cheese

Bone, glass, steel, and wood Metal, glass, and plastic pieces Glass fragment Bone fragment Plastic pieces Rock and aluminum plate Metal rod, metal, and glass fragment Copper and aluminum rods Wood, rubber, and glass

€ m & Luukkala, 2001) (Hæeggstro (Zhao et al., 2003) (Zhao et al., 2006) (Correia et al., 2008) (Leemans & Destain, 2009) (Meftah & Mohd Azimin, 2012) (Cho & Irudayaraj, 2003) (Ho et al., 2007) (Pallav, Hutchins, & Gan, 2009)

Terahertz

Chocolate Flour sample

Stone, glass, and plastic fragments Aluminum foil, metallic cubes, cubic stones, grasshopper, and mealworms Aluminium, granite, and insects Dried fish and metal fragment Crickets species Insect, polymer, and metals Metal razor and rubber fragments Teflon and Polyvinyl chloride (PVC) Caterpillar

€ rdens & Koch, 2008) (Jo (Kim et al., 2012)

Near-infrared (NIR) spectroscopy

Noodle Crackers product Instant noodles Powdered milk Powdered milk Powder Chocolate product

(Lee et al., 2012) (Han et al., 2012) (Ok et al., 2012) (Ok et al., 2014) (Lee & Lee, 2014) (Ikari et al., 2014) (Yu et al., 2015)

foreign body based on the reconstructed images. The combination of

techniques. For example, the combination of conventional X-ray and

noninvasive techniques could be yield an interesting output as one

terahertz technique. Conventional X-ray has limitation in imaging and

technique could overcome the limitations inherent in other

analysis the low-density materials such as rubber and plastic.

16 of 20

|

TA BL E 3

MOHD KHAIRI

ET AL.

The advantages and disadvantages of sensor techniques for detecting foreign bodies in food

Sensor type

Sensor principle

Advantages

Disadvantages

X-ray

Electromagnetic radiation

 Nondestructive measurement  High imaging resolution.

 High cost  High power usage

Thermal imaging

Thermal emission

 Noncontact measurement  No harmful radiation  Able to operates in real-time

 Temperature interference from other surfaces

Near-infrared (NIR) spectroscopy

Optical absorption and reflection

 Nonionizing technique  Able to penetrate through air gaps within the food materials

 Depends on reference methods for calibration purpose

Hyperspectral imaging (HSI)

Optical absorption and reflection

 Provide spatial and spectral information.  Sensitive to minor components.

 Requires a lengthy time for pre-processing of the data and classification

Ultrasonic

Sound waves transmission, reflection and refraction

 Nondestructive technique  Fast response  Able to perform high-velocity measurement

 Limited to a single point measurement  Not appropriate in air medium

Terahertz

Electromagnetic radiation

 Low photon energy  Able to penetrate various materials

 Highly attenuated in water medium  Limited spatial resolution  Low signal-to-noise ratio

Therefore, the combination with terahertz; a method which is recog-

which will be the significant trend in its future application in food safety

nized that can penetrate various materials could be great innovation

and quality aspect.

and contribution in foreign body detection system. Developing a low cost noninvasive technique also is very important issue in food safety

ACK NOWLE DGME NT

and quality area as the current cost for commercialize of noninvasive with imaging technique is still very high.

The authors would like to acknowledge the assistance of the Ministry of Higher Education Malaysia under the MyBrain15 program and

4 | CONCLUSIONS

Universiti Teknologi Malaysia for providing the research grants 15H85 and 4J255 which enabled this research to be carried out.

This review summarized the applications of noninvasive techniques for detecting foreign bodies in food products which included X-ray, thermal

CON FLICT OF INT E RE ST

imaging, near-infrared spectroscopy, hyperspectral imaging, ultrasonic,

The authors hereby declare that they have no conflict of interest.

and terahertz. The sources of contamination of food in production stages are also outlined and encompassed in this article. It was revealed that these noninvasive techniques had the capability to inspect the presence of foreign bodies in several types of foods. Several types of foreign

ORC ID Mohd Taufiq Mohd Khairi

http://orcid.org/0000-0003-0908-141X

bodies also been investigated to recognize the performances and limitations of the techniques. In addition, advantages and disadvantages of

R EFE R ENC E S

each noninvasive methods were also presented. In general, the use of

Abbas, S., Hayat, K., Karangwa, E., Bashari, M., & Zhang, X. (2013). An overview of ultrasound-assisted food-grade nanoemulsions. Food Engineering Reviews, 5(3), 139–157.

certain inspection technology is related to the characteristics of the food and the specific purpose of the inspection. The sensors should be selected based on suitability such as cost, hardware size, risk, detection capability, and environmental situation. The sensors which have been reviewed in this article are widely used for real time processing applications. Understanding the best way to evaluate food product is important to the food manufacturer. In fact, the selection of sensor technique for foreign bodies detection is highly reliant on the type of food that they produced. Techniques such as NIR, hyperspectral, and thermal imaging are suited to distinguish between food (e.g., fruits, cereals, and meats) and foreign body through the external attributes such as color recognition and physical size. Conversely, X-rays, ultrasound, and terahertz are more appropriate to observe the internal attributes of the food by exploiting the high penetration capabilities of these techniques. Some criteria are essentially being considered; (1) speed of system, (2) foreign body classification, (3) combination technique, and (4) low-cost system

 , A., Chavez, J. A., García, Alava, J. M., Sahi, S. S., García-Alvarez, J., Turo M. J., & Salazar, J. (2007). Use of ultrasound for the determination of flour quality. Ultrasonics, 46(3), 270–276. Amon, F., Hamins, A., Bryner, N., & Rowe, J. (2008). Meaningful performance evaluation conditions for fire service thermal imaging cameras. Fire Safety Journal, 43(8), 541–550. Ashley, B. C., Birchfield, P. T., Chamberlain, B. V., Kotwal, R. S., McClellan, S. F., Moynihan, S., . . . Au, W. W. (2004). Health concerns regarding consumption of irradiated food. International Journal of Hygiene and Environmental Health, 207(6), 493–504. Awad, T. S., Moharram, H. A., Shaltout, O. E., Asker, D., & Youssef, M. M. (2012). Applications of ultrasound in analysis, processing and quality control of food: A review. Food Research International, 48(2), 410–427. Bandyopadhyay, O., Biswas, A., & Bhattacharya, B. B. (2016). Long-bone fracture detection in digital X-ray images based on digital-geometric techniques. Computer Methods and Programs in Biomedicine, 123, 2–14.

MOHD KHAIRI

ET AL.

Bansal, S., Singh, A., Mangal, M., Mangal, A. K., & Kumar, S. (2017). Food adulteration: Sources, health risks, and detection methods. Critical Reviews in Food Science and Nutrition, 57(6), 1174–1189.

|

17 of 20

objects in food using multi-modal X-ray imaging. Food Control, 67, 39–47.

Baranowski, P., Mazurek, W., Wozniak, J., & Majewska, U. (2012). Detection of early bruises in apples using hyperspectral data and thermal imaging. Journal of Food Engineering, 110(3), 345–355.

 mez-Ullate, Y., Resa, P., Iglesias, Elvira, L., Sampedro, L., Matesanz, J., Go J. R., . . . de Espinosa, F. M. (2005). Non-invasive and non-destructive ultrasonic technique for the detection of microbial contamination in packed UHT milk. Food Research International, 38(6), 631–638.

Bhaskaracharya, R. K., Kentish, S., & Ashokkumar, M. (2009). Selected applications of ultrasonics in food processing. Food Engineering Reviews, 1(1), 31–49.

FAO/IAEA/WHO. (1999). High-dose irradiation: Wholesomeness of food irradiated with doses above 10 kGy. World Health Organization, 890, 1–197.

Bhuvaneswari, K., Fields, P. G., White, N. D. G., Sarkar, A. K., Singh, C. B., & Jayas, D. S. (2011). Image analysis for detecting insect fragments in semolina. Journal of Stored Products Research, 47(1), 20–24.

Freitas, S., Hielscher, G., Merkle, H. P., & Gander, B. (2006). Continuous contact- and contamination-free ultrasonic emulsification - A useful tool for pharmaceutical development and production. Ultrasonics Sonochemistry, 13(1), 76–85.

Bukowska-Belniak, B., Lesniak, A., Kiełkowski, P., & Michalski, R. (2010). Detection of foreign bodies in comestible product using sequence of low contrast thermographic images. Paper present at the Proceedings of the 10th International Conference on Quantitative InfraRed Thermography (pp. 475–480), Quebec City, Canada.

Garboczi, E. J. (2002). Three-dimensional mathematical analysis of particle shape using X-ray tomography and spherical harmonics: Application to aggregates used in concrete. Cement and Concrete Research, 32(10), 1621–1638.

Butz, P., Hofmann, C., & Tauscher, B. (2006). Recent developments in noninvasive techniques for fresh fruit and vegetables internal quality analysis. Journal of Food Science, 70(9), R131–R141.

Ginesu, G., Giusto, D. D., Märgner, V., & Meinlschmidt, P. (2004). Detection of foreign bodies in food by thermal image processing. IEEE Transactions on Industrial Electronics, 51(2), 480–490.

Chandrapala, J., & Leong, T. (2015). Ultrasonic processing for dairy applications: Recent advances. Food Engineering Reviews, 7(2), 143–158.

Gorenflo, S., Tauer, U., Hinkov, I., Lambrecht, A., Buchner, R., & Helm, H. (2006). Dielectric properties of oil-water complexes using terahertz transmission spectroscopy. Chemical Physics Letters, 421(4–6), 494–498.

Chandrapala, J., Oliver, C., Kentish, S., & Ashokkumar, M. (2012a). Ultrasonics in food processing. Ultrasonics Sonochemistry, 19(5), 975–983. Chandrapala, J., Oliver, C., Kentish, S., & Ashokkumar, M. (2012b). Ultrasonics in food processing – Food quality assurance and food safety. Trends in Food Science & Technology, 26(2), 88–98.

Gowen, A. A., & O’Donnell, C. P. (2013). Near infrared hyperspectral imaging for foreign body detection and identification in food processing. Spectroscopy Europe, 25, 6–11.

Chemat, F., Zill-e-Huma., & Khan, M. K. (2011). Applications of ultrasound in food technology: Processing, preservation and extraction. Ultrasonics Sonochemistry, 18, 813–835.

Gowen, A. A., O’Sullivan, C., & O’Donnell, C. P. (2012). Terahertz time domain spectroscopy and imaging: Emerging techniques for food process monitoring and quality control. Trends in Food Science and Technology, 25(1), 40–46.

Chen, X., Jing, H., Tao, Y., & Cheng, X. (2005). Real-time image analysis for nondestructive detection of metal sliver in packed food. Proceedings of SPIE, 5996, 120–129.

Gowen, A. A., Tiwari, B. K., Cullen, P. J., McDonnell, K., & O’Donnell, C. P. (2010). Applications of thermal imaging in food quality and safety assessment. Trends in Food Science and Technology, 21(4), 190–200.

Chen, Q., Zhang, C., Zhao, J., & Ouyang, Q. (2013). Recent advances in emerging imaging techniques for non-destructive detection of food quality and safety. Trends in Analytical Chemistry, 52, 261–274.

Graves, M., Smith, A., & Batchelor, B. (1998). Approaches to foreign body detection in foods. Trends in Food Science & Technology, 9(1), 21–27.

Cho, B. K., & Irudayaraj, J. M. K. (2003). Foreign object and internal disorder detection in food materials using noncontact ultrasound imaging. Journal of Food Science, 68(3), 967–974. Chua, H. S., Obradovic, J., Haigh, A. D., Upadhya, P. C., Hirsch, O., Crawley, D., . . . Linfield, E. H. (2005). Terahertz time-domain spectroscopy of crushed wheat grain. Paper presented at the IEEE MTT-S International Microwave Symposium Digest (pp. 1–4), Long Beach, CA. Correia, L. R., Mittal, G. S., & Basir, O. A. (2008). Ultrasonic detection of bone fragment in mechanically deboned chicken breasts. Innovative Food Science and Emerging Technologies, 9(1), 109–115. Debastiani, R., Simon, R., Batchelor, D., Dellagustin, G., Baumbach, T., & Fiederle, M. (2016). Synchrotron-based scanning macro-X-ray fluorescence applied to fragments of Roman mural paintings. Microchemical Journal, 126, 438–445.  Díaz, R., Cervera, L., Fenollosa, S., Avila, C., & Belenguer, J. (2011). Hyperspectral system for the detection of foreign bodies in meat products. Paper presented at the Procedia Engineering Eurosensors XXV (pp. 313–316), Athens, Greece. Edwards, M. (2004). Detecting foreign bodies in food. Cambridge, England: Woodhead Publishing Limited. Edwards, M. C., & Stringer, M. F. (2007). Observations on patterns in foreign material investigations. Food Control, 18(7), 773–782.  ttir, H., Emerson, M. J., Clemmensen, L. H., Scherer, K., Willer, Einarsdo K., Bech, M., . . . Pfeiffer, F. (2016). Novelty detection of foreign

Guillet, J. P., Recur, B., Frederique, L., Bousquet, B., Canioni, L., ManekHonninger, I., . . . Mounaix, P. (2014). Review of terahertz tomography techniques. Journal of Infrared, Millimeter, and Terahertz Waves, 35(4), 382–411. €m, E., & Luukkala, M. (2001). Ultrasound detection and identifiHæggstro cation of foreign bodies in food products. Food Control, 12(1), 37–45. Haff, R. P., & Toyofuku, N. (2008). X-ray detection of defects and contaminants in the food industry. Sensing and Instrumentation for Food Quality and Safety, 2(4), 262–273. Han, S. T., Park, W. K., Ahn, Y. H., Lee, W. J., & Chun, H. S. (2012). Development of a compact sub-terahertz gyrotron and its application to T-ray real-time imaging for food inspection. Paper presented at the 37th International Conference on Infrared, Millimeter, and Terahertz Waves (pp. 6–7), Wollongong, Australia. Hauptmann, P., Hoppe, N., & Puttmer, A. (2002). Application of ultrasonic sensors in the process industry. Measurement Science and Technology, 13(8), R73–R83. Henning, B., & Rautenberg, J. (2006). Process monitoring using ultrasonic sensor systems. Ultrasonics, 44, e1395–e1399. Ho, K. S., Billson, D. R., & Hutchins, D. A. (2007). Inspection of drinks cans using non-contact electromagnetic acoustic transducers. Journal of Food Engineering, 80(2), 431–444. Huang, M., Kim, M. S., Chao, K., Qin, J., Mo, C., Esquerre, C., . . . Zhu, Q. (2016). Penetration depth measurement of near-infrared hyperspectral imaging light for milk powder. Sensors, 16(4), 441–411.

18 of 20

|

MOHD KHAIRI

ET AL.

Huang, H., Liu, L., & Ngadi, M. O. (2014). Recent developments in hyperspectral imaging for assessment of food quality and safety. Sensors, 14(4), 7248–7276.

Lahiri, B. B., Bagavathiappan, S., Jayakumar, T., & Philip, J. (2012). Medical applications of infrared thermography: A review. Infrared Physics and Technology, 55(4), 221–235.

Hyman, F. N., Klontz, K. C., & Tollefson, L. (1993). Food and drug administration surveillance of the role of foreign objects in foodborne injuries. Public Health Reports, 108, 54–59.

Lee, Y.-K., Choi, S.-W., Han, S.-T., Woo, D. H., & Chun, H. S. (2012). Detection of foreign bodies in foods using continuous wave terahertz imaging. Journal of Food Protection, 75(1), 179–183.

Ikari, T., Takahashi, N., Fukasawa, R., & Duling, I. (2014). Non-metallic foreign matter detection in powder using terahertz pulse. Paper presented at the 39th International Conference on Infrared, Millimeter, and Terahertz waves (pp. 1–2), Tucson, AZ.

Lee, W.-H., & Lee, W. (2014). Food inspection system using terahertz imaging. Microwave and Optical Technology Letters, 56(5), 1211–1214. Leemans, V., & Destain, M.-F. (2009). Ultrasonic internal defect detection in cheese. Journal of Food Engineering, 90(3), 333–340.

James, B. (2005). Foreign body contamination of food - Scanning electron microscopy and energy dispersive spectroscopy as tools for identification. International Journal of Food Engineering, 1, 1–15.

Leong, T. S. H., Wooster, T. J., Kentish, S. E., & Ashokkumar, M. (2009). Minimising oil droplet size using ultrasonic emulsification. Ultrasonics Sonochemistry, 16(6), 721–727.

Jepsen, P. U., Møller, U., & Merbold, H. (2007). Investigation of aqueous alcohol and sugar solutions with reflection terahertz time-domain spectroscopy. Optics Express, 15(22), 14717–14737.

Lewis, D. F. (1993). A tutorial and comprehensive bibliography on the identification of foreign bodies found in food. Food Structure, 12(3), 365–378.

Jha, S. N., Narsaiah, K., Basediya, A. L., Sharma, R., Jaiswal, P., Kumar, R., & Bhardwaj, R. (2011). Measurement techniques and application of electrical properties for nondestructive quality evaluation of foods - A review. Journal of Food Science and Technology, 48(4), 387–411. Jiang, L., Zhu, B., Rao, X., Berney, G., & Tao, Y. (2007). Discrimination of black walnut shell and pulp in hyperspectral fluorescence imagery using Gaussian kernel function approach. Journal of Food Engineering, 81(1), 108–117. Jiusheng, L. (2010). Optical parameters of vegetable oil studied by terahertz time-domain spectroscopy. Applied Spectroscopy, 64(2), 231– 234. € rdens, C., & Koch, M. (2008). Detection of foreign bodies in chocolate Jo with pulsed terahertz spectroscopy. Optical Engineering, 47(3), 037003. Kamruzzaman, M., Elmasry, G., Sun, D. W., & Allen, P. (2011). Application of NIR hyperspectral imaging for discrimination of lamb muscles. Journal of Food Engineering, 104(3), 332–340. Kateb, B., Yamamoto, V., Yu, C., Grundfest, W., & Gruen, J. P. (2009). Infrared thermal imaging: A review of the literature and case report. NeuroImage, 47, T154–T162. Kentish, S., & Feng, H. (2014). Applications of power ultrasound in food processing. Annual Review of Food Science and Technology, 5(1), 263– 284. Kim, G. J., Kim, J.-I., Jeon, S.-G., Kim, J., Park, K.-K., & Oh, C.-H. (2012). Enhanced continuous-wave terahertz imaging with a horn antenna for food inspection. Journal of Infrared, Millimeter, and Terahertz Waves, 33(6), 657–664. Kim, K.-B., Lee, S., Kim, M.-S., & Cho, B.-K. (2009). Determination of apple firmness by nondestructive ultrasonic measurement. Postharvest Biology and Technology, 52(1), 44–48. Knorr, D., Zenker, M., Heinz, V., & Lee, D.-U. (2004). Applications and potential of ultrasonics in food processing. Trends in Food Science & Technology, 15(5), 261–266.

Li, F., Liu, Z., Sun, T., Ma, Y., & Ding, X. (2015). Confocal threedimensional micro X-ray scatter imaging for non-destructive detecting foreign bodies with low density and low-Z materials in food products. Food Control, 54, 120–125. Li, J. L., Sun, D. W., & Cheng, J. H. (2016). Recent advances in nondestructive analytical techniques for determining the total soluble solids in fruits: A review. Comprehensive Reviews in Food Science and Food Safety, 15(5), 897–911. Lim, K. S., & Barigou, M. (2004). X-ray micro-computed tomography of cellular food products. Food Research International, 37(10), 1001–1012. Liu, Y., Pu, H., & Sun, D.-W. (2017). Hyperspectral imaging technique for evaluating food quality and safety during various processes: A review of recent applications. Trends in Food Science & Technology, 69, 25. Lu, Z., & Peng, N. (2013). Dual view x-ray inspection system for foreign objects detection in canned food. Paper presented at the Proceedings of SPIE 8788, Optical Measurement Systems for Industrial Inspection VIII (pp. 1–9), Munich, Germany. Mangus, D. L., Sharda, A., & Zhang, N. (2016). Development and evaluation of thermal infrared imaging system for high spatial and temporal resolution crop water stress monitoring of corn within a greenhouse. Computers and Electronics in Agriculture, 121, 149–159. Manley, M. (2014). Near-infrared spectroscopy and hyperspectral imaging: Non-destructive analysis of biological materials. Chemical Society Reviews, 43(24), 8200–8214. Marsh, R. A., & Angold, R. E. (2004). Identifying potential sources of foreign bodies in the supply chain. In M. C. Edwards (Ed.), Detecting foreign bodies in food (pp. 3–13). Cambridge, England: Woodhead Publishing Limited. Mathiassen, J. R., Misimi, E., Bondø, M., Veliyulin, E., & Østvik, S. O. (2011). Trends in application of imaging technologies to inspection of fish and fish products. Trends in Food Science and Technology, 22(6), 257–275. Mcclements, D. J., & Sundaram, G. (1997). Ultrasonic characterization of foods and drinks: Principles, methods and applications. Critical Reviews in Food Science and Nutrition, 37(1), 1–46.

Kong, W., Zhang, C., Liu, F., Nie, P., & He, Y. (2013). Rice seed cultivar identification using near-infrared hyperspectral imaging and multivariate data analysis. Sensors, 13(7), 8916–8927.

McClure, W. F., Moody, D., Stanfield, D. L., & Kinoshita, O. (2002). Hand-held NIR Spectrometry. Part II: An economical no-moving parts spectrometer for measuring chlorophyll and moisture. Applied Spectroscopy, 56(6), 720–724.

Kuzy, J., & Li, C. (2017). A pulsed thermographic imaging system for detection and identification of cotton foreign matter. Sensors, 17(3), 518–515.

Meftah, H., & Mohd Azimin, E. (2012). Detection of foreign bodies in canned foods using ultrasonic testing. International Food Research Journal, 19, 543–546.

Kwon, J. S., Lee, J. M., & Kim, W. Y. (2008). Real-time detection of foreign objects using x-ray imaging for dry food manufacturing line. Paper presented at the Proceedings of the International Symposium on Consumer Electronics (pp. 1–4), Vilamoura, Portugal.

Meinel, F. G., Schwab, F., Yaroshenko, A., Velroyen, A., Bech, M., Hellbach, K., . . . Nikolaou, K. (2014). Lung tumors on multimodal radiographs derived from grating-based X-ray imaging - A feasibility study. Physica Medica, 30(3), 352–357.

MOHD KHAIRI

ET AL.

Meinlschmidt, P., & Maergner, V. (2002). Detection of foreign substances in food using thermography. Paper presented at the Proceedings of SPIE 4710, Thermosense XXIV (pp. 565–571), Bellingham, WA. Mery, D., Lillo, I., Loebel, H., Riffo, V., Soto, A., Cipriano, A., & Aguilera, J. M. (2011). Automated fish bone detection using X-ray imaging. Journal of Food Engineering, 105(3), 485–492. Moghaddam, T. M., Razavi, S. M. A., & Taghizadeh, M. (2013). Applications of hyperspectral imaging in grains and nuts quality and safety assessment: A review. Journal of Food Measurement and Characterization, 7(3), 129–140. Mohammadi, V., Ghasemi-Varnamkhasti, M., Ebrahimi, R., & Abbasvali, M. (2014). Ultrasonic techniques for the milk production industry. Measurement, 58, 93–102. Mongenot, N., Charrier, S., & Chalier, P. (2000). Effect of ultrasound emulsification on cheese aroma encapsulation by carbohydrates. Journal of Agricultural and Food Chemistry, 48(3), 861–867. Morita, K., Ogawa, Y., Thai, C. N., & Tanaka, F. (2003). Soft X-ray image analysis to detect foreign materials in foods. Food Science and Technology Research, 9(2), 137–141. Morrison, D. S., & Abeyratne, U. R. (2014). Ultrasonic technique for nondestructive quality evaluation of oranges. Journal of Food Engineering, 141, 107–112. Mousavi, R., Miri, T., Cox, P. W., & Fryer, P. J. (2005). A novel technique for ice crystal visualization in frozen solids using X-ray micro-computed tomography. Journal of Food Science, 70(7), e437–e442. Nicolaï, B. M., Defraeye, T., Ketelaere, B., De, Herremans, E., Hertog, M. L. A. T. M., Saeys, W., . . . Verboven, P. (2014). Nondestructive measurement of fruit and vegetable quality. Annual Review of Food Science and Technology, 5(1), 285–312. Nielsen, M. S., Lauridsen, T., Christensen, L. B., & Feidenhans’l, R. (2013). X-ray dark-field imaging for detection of foreign bodies in food. Food Control, 30(2), 531–535. Niemeyer, W. D. (2015). SEM/EDS analysis for problem solving in the food industry. Paper presented at the Proceedings of SPIE 9636, Scanning Microscopies (pp. 1–9), California, CA. Ok, G., Choi, S. W., Park, K. H., & Chun, H. S. (2012). Foreign object detection by sub-terahertz quasi-Bessel beam imaging. Sensors, 13(1), 71–85. Ok, G., Kim, H. J., Chun, H. S., & Choi, S. W. (2014). Foreign-body detection in dry food using continuous sub-terahertz wave imaging. Food Control, 42, 284–289. Olsen, A. R. (1998). Regulatory action criteria for filth and other extraneous materials. I. Review of hard or sharp foreign objects as physical hazards in food. Regulatory Toxicology and Pharmacology, 28(3), 181–189. Orina, I., Manley, M., & Williams, P. J. (2017). Non-destructive techniques for the detection of fungal infection in cereal grains. Food Research International, 100, 74–86. Osborne, B., Fearn, T., & Hindle, P. (1993). Practical NIR spectroscopy with applications in food and beverage analysis (2nd ed.). London, England: Longman Scientific & Technical. Pallav, P., Diamond, G. G., Hutchins, D. A., Green, R. J., & Gan, T. H. (2009). A near-infrared (NIR) technique for imaging food materials. Journal of Food Science, 74(1), E23–E33. Pallav, P., Hutchins, D. A., & Gan, T. H. (2009). Air-coupled ultrasonic evaluation of food materials. Ultrasonics, 49(2), 244–253. Parasoglou, P., Parrott, E. P. J., Zeitler, J. A., Rasburn, J., Powell, H., Gladden, L. F., & Johns, M. L. (2009). Quantitative moisture content detection in food wafers. Paper presented at the 34th International Conference on Infrared, Millimeter, and Terahertz Waves (pp. 1–2), Busan, Korea.

|

19 of 20

Peariso, D. (2006). Preventing foreign material contamination of foods. Iowa, IA: Blackwell Publishing. Phetchalern, S., Tashima, H., Ishii, Y., Ishiyama, T., Arai, S., & Fukuda, M. (2014). Near-infrared imaging equipment that detects small organic substances in thick foods. Paper presented at the Proceedings of SPIE Current Developments in Lens Design and Optical Engineering XV (pp. 1–6), California, CA. Qin, J., Ying, Y., & Xie, L. (2013). The detection of agricultural products and food using terahertz spectroscopy: A review. Applied Spectroscopy Reviews, 48(6), 439–457.  s, E., García-Reguero, Redo-Sanchez, A., Salvatella, G., Galceran, R., Roldo J.-A., Castellari, M., & Tejada, J. (2011). Assessment of terahertz spectroscopy to detect antibiotic residues in food and feed matrices. The Analyst, 136(8), 1733–1738.  , A., Ch Salazar, J., Turo avez, J. A., & García, M. J. (2004). Ultrasonic inspection of batters for on-line process monitoring. Ultrasonics, 42 (1–9), 155–159. Schoeman, L., Williams, P., Plessis, A., & Du Manley, M. (2016). X-ray micro-computed tomography (lCT) for non-destructive characterisation of food microstructure. Trends in Food Science and Technology, 47, 10–24. Senni, L., Burrascano, P., & Ricci, M. (2016). Multispectral laser imaging for advanced food analysis. Infrared Physics and Technology, 77, 179– 192. Senni, L., Ricci, M., Palazzi, A., Burrascano, P., Pennisi, P., & Ghirelli, F. (2014). On-line automatic detection of foreign bodies in biscuits by infrared thermography and image processing. Journal of Food Engineering, 128, 146–156. Shanmugam, A., Chandrapala, J., & Ashokkumar, M. (2012). The effect of ultrasound on the physical and functional properties of skim milk. Innovative Food Science & Emerging Technologies, 16, 251–258. Siche, R., Vejarano, R., Aredo, V., Velasquez, L., Saldana, E., & Quevedo, R. (2016). Evaluation of food quality and safety with hyperspectral imaging (HSI). Food Engineering Reviews, 8(3), 306–322. Sigfusson, H., Ziegler, G. R., & Coupland, J. N. (2004). Ultrasonic monitoring of food freezing. Journal of Food Engineering, 62(3), 263–269. Stajnko, D., Lakota, M., & Hočevar, M. (2004). Estimation of number and diameter of apple fruits in an orchard during the growing season by thermal imaging. Computers and Electronics in Agriculture, 42(1), 31–42. Stringer, M. F., & Hall, M. N. (2007). A generic model of the integrated food supply chain to aid the investigation of food safety breakdowns. Food Control, 18(7), 755–765. Sugiyama, T., Sugiyama, J., Tsuta, M., Fujita, K., Shibata, M., Kokawa, M., . . . Sagara, Y. (2010). NIR spectral imaging with discriminant analysis for detecting foreign materials among blueberries. Journal of Food Engineering, 101(3), 244–252. Sun, T., Liu, Z., Wang, G., Ma, Y., Peng, S., Sun, W., . . . Ding, X. (2014). Application of confocal X-ray fluorescence micro-spectroscopy to the investigation of paint layers. Applied Radiation and Isotopes, 94, 109–112. Tashima, H., Genta, T., Ishii, Y., Ishiyama, T., Arai, S., & Fukuda, M. (2013). Near-infrared imaging system for detecting small organic foreign substances in foods. Paper presented at the Proceedings of SPIE 8841, Current Developments in Lens Design and Optical Engineering XIV (pp. 1–8), California, CA. Tauxe, R. V. (2001). Food safety and irradiation: Protecting the public from foodborne infections. Emerging Infectious Diseases, 7(7), 516– 521. Trafialek, J., Kaczmarek, S., & Kolanowski, W. (2016). The risk analysis of metallic foreign bodies in food product. Journal of Food Quality, 0, 1– 10.

20 of 20

|

Tsuta, M., Takao, T., Sugiyama, J., Wada, Y., & Sagara, Y. (2006). Foreign substance detection in blueberry fruits by spectral imaging. Food Science and Technology Research, 12(2), 96–100. Vadivambal, R., & Jayas, D. S. (2011). Applications of thermal imaging in agriculture and food industry - A review. Food and Bioprocess Technology, 4(2), 186–199. Wang, L. B., Frost, J. D., Voyiadjis, G. Z., & Harman, T. P. (2003). Quantification of damage parameters using X-ray tomography images. Mechanics of Materials, 35(8), 777–790. Wang, W., & Paliwal, J. (2007). Near-infrared spectroscopy and imaging in food quality and safety. Sensing and Instrumentation for Food Quality and Safety, 1(4), 193–207. Woh, P. Y., Thong, K. L., Behnke, J. M., Lewis, J. W., & Mohd Zain, S. N. (2016). Evaluation of basic knowledge on food safety and food handling practices amongst migrant food handlers in Peninsular Malaysia. Food Control, 70, 64–73. Xiong, Z., Sun, D.-W., Zeng, X.-A., & Xie, A. (2014). Recent developments of hyperspectral imaging systems and their applications in detecting quality attributes of red meats: A review. Journal of Food Engineering, 132, 1–13. Yoon, S. C., Lawrence, K. C., Smith, D. P., Park, B., & Windham, W. R. (2008). Bone fragment detection in chicken breast fillets using transmittance image enhancement. Transactions of the American Society of Agricultural and Biological Engineers, 51, 331–339.

MOHD KHAIRI

ET AL.

Yu, X., Endo, M., Ishibashi, T., Shimizu, M., Kusanagi, S., Nozokido, T., & Bae, J. (2015). Orthogonally polarized terahertz wave imaging with real-time capability for food inspection. Paper presented at the AsiaPacific Microwave Conference (pp. 1–3), Nanjing, China. Yuefang, H., & Hongjian, Z. (2010). Qualitative and quantitative detection of pesticides with terahertz time-domain spectroscopy. IEEE Transactions on Microwave Theory and Techniques, 58, 2064–2070. Zhao, B., Basir, O. A., & Mittal, G. S. (2003). Detection of metal, glass and plastic pieces in bottled beverages using ultrasound. Food Research International, 36(5), 513–521. Zhao, B., Yang, P., Basir, O. A., & Mittal, G. S. (2006). Ultrasound based glass fragments detection in glass containers filled with beverages using neural networks and short time Fourier transform. Food Research International, 39(6), 686–695. Zheng, L., & Sun, D.-W. (2006). Innovative applications of power ultrasound during food freezing processes - A review. Trends in Food Science & Technology, 17(1), 16–23.

How to cite this article: Mohd Khairi MT, Ibrahim S, Md Yunus MA, Faramarzi M. Noninvasive techniques for detection of foreign bodies in food: A review. J Food Process Eng. 2018;e12808. https://doi.org/10.1111/jfpe.12808

Suggest Documents