The Journal of The Textile Institute Vol. 101, No. 1, January 2010, 2-7
Taylor 6i Francis IjyhH ^Franui Group
Evaluation of diamond bar patterns on fabric surface using an image analysis technique M. Ghane*, S.A. Hosseini Ravandi and M. Moezzi Department of Textile Engineering, Isfahan University of Technology, Isfahan H4! 56-8311 /. ¡ran {Received 20 Februaiy 2008:finalversion received 8 May 2008) The main aim of this work ¡s to study of the effect of yarn periodic irregularities on the plain weave fabric appearance using an image analysis method. The pattern of fabric faults was simulated using a prepared computer program. In the experimental stage, two different types of yam were prepared. Thefirstyam was produeed with a certain wavelength and the second yam was prepared with a strong periodic irregularity using an open-end spinning machine. The wavelength of the open-end yarn fault was measured by using an Uster 4 automatic evenness tester. For practical examination, the yams were used as weft in a shuttle loom machine and two types of fabrics were produced. Using image processing, the number and size of the diamond bar pattems, and also the theoretical wavelength of the weft yam, were calculated. The wavelengths calculated theoretically were compared with the actual wavelengths of the weft yams. The results showed an acceptable accuracy of the method. Keywords: yam periodic faults; wavelength; fabric effective width; fabric appearance; diamond bar Introduction In recent years, the improvement of computer technology has been positively effective on different parts of textile industry such as on-line monitoring and automation, but some of the textile processes such as identifying the yarn periodic defects in fabric and assessing fabric surface, are carried out manually leading to fatigue, low speed, less accuracy, etc. Yam defects have a dominant efFect on the fabric properties such as luster, handle., and esthetic. Some factors, i.e. weft and warp density, width of fabric, fabric weave, wavelength of weft yarn faults, weft and warp yam contraction, and the kind of take-up mechanism have been known as the source of surface irregularities in the final fabrics. Many researchers have studied the effect of different kinds of yam irregularities on fabric surfaces (Furter, 1982; Keisokki Kogyo Co. Ltd., 1986). In some cases, the relation between the wavelength of defects and the shape and pattern of defects on the fabric have been considered (Catling, 1958; Seyam & El-Shiekh, 1990). The application of the Fourier transform and an image analysis technique on the evaluation of the fabric surface was also studied (Bugao, 1996; Cardamone, 2002). Due to the periodic nature of pattems on the fabric surface, the use of the Fourier transform is feasible (Sakaguchi, Wen, Matsumoto, Toriumi, & Kim, 2001). In this research, the construction parameters of the weft yam, i.e. wavelength irregularity, yarn count, and the fabric
'Corresponding author. Email:
[email protected] ISSN 0040-5000 print / ISSN 1754-2340 online Copyright © 2010 The Textile Institute DOr: HI, 108l)/()04()500[)802190703 h ttp : ' /w w w. i n Ib rmawo rl d. cü m
parameters; weft density and fabric width, were used to predict and evaluate the number and size of the diamond bar pattems on the fabric surface. On the other hand, the image analysis technique was applied to the fabric surface and the periodic variation of the weft yam was calculated theoretically and compared with the actual value of the weft yarn irregularities. Definitions Considering a fabric produced by a shuttle loom weaving machine, the relation between fabric width and periodic wavelength of the weft yam is defined as follows:
(1) where: W = effective fabric width, X = periodic wavelength of weft yarn, P ^ a real value larger than 2, and r = a value between —0.5 and -1-0.5. It can be shown that with different values of r, different kinds of pattems will appear on the fabric surface. When the r value is nonzero between [-0.5, +0.5], a kind of pattem called diamond bar is formed in the weft-wise and warp-wise of the fabric surface. The accumulation of the thick places of the weft yarn form thick diamonds and
The Journal of The Textile Institute
ler
Figure 1. Spectrogram of the weft yarn used in fabric 2.
the thin places of the weft yarn form thin diamonds. The periodic wavelength of the weft yarn fault is (Foster, 1952): PBXW
(2)
(PB
where; RA is the number of repeated patterns in fabric width, including one thick and one thin diamond, and Pß is the number of inserted picks in relation to each repeated pattern or number of picks from the center of one thick diamond to the next one. Discrete fou Her transform Two-dimensional Fourier transforms are a simple method to transform the images to frequency domain. The discrete Fourier transform pairs when images are sampled in a squared array are given by the following equations: I N-l
N-\
F (it. v) = -—T 2_[ y j /(-Ï. .v)exp[—277r (ux + vy) /N] .t=()
for», t' = 0. 1.2
1 y)=—
f(x.
N-l
>•=()
N - 1. and
(3)
N-l
Figure 2. Captured image of fabric 1.
(6) The periodic nature of the fabric surface permits the use of the Fourier power spectrum method to consider fabric appearance (Hosseini Ravandi & Toriumi. 1995; Huang, Liu, & Wen, 2000). The two-dimensional power spectrum function was calculated as follows:
in (ux -\- vy)/N'\ 11=0 v-0
:,>' = 0. 1 , 2 , . . . , / V - 1.
The sampling increments in the spatial and frequency domain are related by Equations (5) and (6):
AH
=
NAx
(5)
Table 1. The fabric specifications. Weft density Ends/cm 14
Warp density Ends/cm 22
Reed number
f o r « , i> = 0, 1.2
(4)
Fabric width (cm) 94
Warp yam 20Nc (Cotton/polyester)
A' -
I.
(7)
Experimental Fabric production In this work, two types of fabrics were produced. Dyed weft yarn was used in the first type (fabric 1), and in the second type (fabric 2), a weft yam with a real defect was used. To produce the weft yarn for the first fabric, cotton ring yarn. Ne = 10, was dyed with direct fast blue B2R in such a manner that a 20 cm was dyed and the following 20 cm was not dyed. Consequently, a defect with 40 cm wavelength on the yam surface was achieved. In the second fabric, a real defective weft yam was used. The defective yam was produced by using Autocoro
M. Ghane et al.
Figure 3, Captured image of fabric 2. Figure 6, Residual part of two-dimensional power spectrum of fabric 1 (512 X 512 pixels).
Figure 4. Two-dimensional power spectrum of fabric 1 (512 x 512 pixels). SE8 open-end machine. In the groove of the rotor, a thick point elaborately had been mentioned. The diameter of tbe rotor was 33 mm. Consequently, a periodic defect with 10.4 cm wavelength was produced on the yarn. In a complete wavelength, a length of 9.5 cm was normal and the remaining 0.9 cm was a thick place. Figure 1 shows the spectrograph of the yarn. The specifications of the fabrics were measured using the standard methods and are shown in Table 1.
Figure 5. Two-dimensional power spectrum of fabric 2 (512 x 512 pixels).
Figure 7. Residual part of two-dimensional power spectrum of fabric 2(512 x 512 pixels).
The images of fabrics were captured using a scanner with 100 and 200 pixels per inch (PPI) resolution for fabric I and fabric 2, respectively. Figures 2 and 3 show two typical images of the fabrics.
Figure 8. Typical image of fabric 1 after using inverse Fourier transform.
The Journal of The Textile Institute an image with N x N cells, the intensity of image in the vertical direction was calculated as follows:
(8) j = 1.2
Figure 9. Typical image of fabric 2 after using inverse Fourier transform.
Image processing To obtain the distance between two successive diamonds in the warp direction, first, diserete Fourier transform was used. Figures 4 and 5 show the power spectrum of first and second fabrics, respectively. Then, the low frequency of the power spectrum was filtered. Finally, the inverse Fourier transform was applied, as shown in Figures 6 and 7, and the results obtained are shown in Figures 8 and 9. All parts of the images, except the required pattern, have been deleted.
Measurement of the distance between two successive diamonds Figures 8 and 9 are monochromatic with 256 gray levels. To obtain the distance between two successive diamonds of
20
where: /; : the overall luminance of defective fabric in lengthwise. a¡j : the luminance of each cell of the image matrix. Considering the fabric structure with periodic properties, the distance between two successive diamonds can be determined by using discrete Fourier transform, in this way, the wavelength of the periodic part, i.e. the distance between two successive diamonds can be found as follows: _ N X 25/4 * ~ PPi X K K = N ~ 1
-, 1,0.
The calculated one-dimensional power spectrum has been shown in Figures 10 and 11 (Bcndat, 1986; Brigham. 1988). The dominant peaks observed in Figures 10 and 11 show the distance between two successive diamonds in the weft-wise.
40 60 80 100 120 140 160 180 Distance between two successive diamond bars (mm) ^ X
Figure 10. One-dimensional power spectrum of fabric 1.
(9)
200
M. Ghane et al.
20
40 60 90 100 120 140 160 ISO Distanoe between two successive diamond bars(nnm), A.
200
Figure 11. One-dimensional power spectrum of fabric 2.
Results and discussion The obtained values of the distance between two successive diamond bars are 0.81 and 1.62 mm for fabric 1 and fabric 2, respectively. It is clear that variations in diagrams of intensity of column and row pixels are influenced by the distance between two successive diamonds. Considering the distance between two successive diamonds and weft density, it is possible to determine the number of wefts inserted from the eenter of one diamond to the next one ( P« ). From Equation (9), the wavelength of the weft yarn defect {X) can be calculated. The results are shown in Table 2. The results show a good accuracy of the image processing technique for the fabrics. In the case of fabric 1, two possible reasons can be considered for deviations of the theoretical value of A. from the practical value. First, the penetration of the dye along the yarn due to surface tension during dyeing may lead to inaccuracy of the value of the wavelength. The displacement of pattern is the second reason. As can be seen from Figures 2 and 3, the wavelength of periodic wave of tbe weft yam used in fabric 1 is longer than that of fabric 2. Therefore, less complete pattern repeats exist Table 2. Wavelength of weft yam. Samples Fabric 1 Fabric 2
Weft density W Ends/cm (cm 1 14 14
97 94
f?
PB
k from X from calculation from Uster
2.5 11.4 9 22.7
40.2 10.5
_
40 10.4
across the width of the fabric. This in turn, may lead to less accuracy of the calculations. In the case of fabric 2, as mentioned before, tbe lengths of the thick and thin part of the variation wave are not identical. This leads to smaller size of the diamonds formed by thick places of the weft yarn in comparison to the largersize diamond formed by thin places of the weft yarn. Consequently, the dift"erence between the sizes of tbe diamonds could be a source of error in calculation of the wavelength of the weft yam.
Conclusion The effect of periodic variation of the thickness of the weft yarn on the fabric surface appearance was studied. Using the image processing technique, a method was presented to detect and evaluate this defect. The number and size of the diamond bars were measured and the values of the wavelength of the weft yam were also calculated theoretically. The theoretically calculated values of the wavelengths were then compared with the praetical values of the weft yarn defects obtained from Uster tester 4. The results show that the obtained wavelength from the image processing method has a good and acceptable agreement with the practical values. This method ean be used to predict and calculate tbe number and size of the diamond bar patterns formed on the woven fabric surface caused by the weft yarn periodic faults.
The Journal of The Textile Institute References Bendat, J,S. (1986). Random data analysis and measurement procedures (2nd ed.). New York: Wiley. Brigham, E.O. ( 19S8). The fast Fourier transform and its application. Englcwood Cliffs, NJ: Prentice-Hall. Bugao, X. (1996). An overview of applications of image analysis to objectively evaluate fabric appearance. Textile Chemist and
Cohrist. 28(5). 18-23. Cardamonc, J.M. (2002). Digital image analysis for fabric assesstnent. Textile Research .fournal, 72.906-916. Catling, H.(1958). Some effects of sinusoidal periodic yarn thickness variations on the appearance of woven cloth. Journal of Textile Institute. 49, T234-T246. Foster, R. (1952). Weaving investigations - periodic patterning in íahúc^. Journal of Textile Institute. 43, P742-P754. Furter, R. (1982). Evenness testing in yarn production: Part I. Manchester: The Textile Institute.
Hosseini Ravandi. S.A.. & Toriumi, K. (1995). Fourier transform analysis of plain weave fabric appearance. Textile Research Journal. 65, 676-683. Huang, C.C., Liu. S.C, & Wen. H.Y. (2000). Woven fabric analysis by image processing, parti : Identification of weave pattems. Textile Research Journal. 70, 481-185. Keisokki Kogyo Co., Ltd. (1986). Ba.sic and practice of the evenness testing (pp. 66-69). Meishin-cho, Amagasaki-shi, Hyogo-kcn, Japan: Author. Sakaguchi, A.. Wen. G.H., Matsumoto. Y., Toriumi. K.. & Kim. H. (2001). Image analysis of woven fabric surface irregularity. Textile Research Journal. 71, 666-671. Seyam, A., & El-Shiekh, A. (1990). Mechanics of woven fabric. Part 1 : Theoretical investigation of weave ability limit ofyams with thickness variation. Textile Research Journal, 60, 389404.
Copyright of Journal of the Textile Institute is the property of Taylor & Francis Ltd and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.
