Altered Fingerprint Identification & Classification

Altered Fingerprint Identification & Classification Using SP Detection &Fuzzy Classification Ram Kumar1, Jasvinder Pal Singh2 Gaurav Srivastava3

1 2,3

M.Tech Scholor in R.K.D.F Institute of Technology & Science , Bhopal

Asstant Professor in Computer Science & Engineering in R.K.D.F Institute of Technology & Science , Bhopal {[email protected]; [email protected]; [email protected]}

Abstract: Fingerprint recognition is one of the most commonly used biometric technology. Even if fingerprint temporarily changes (cuts, bruises) it reappears after the finger heals. Criminals started to be aware of this and try to escape the identification systems applying methods from ingenious to very cruel. It is possible to remove, alter or even fake fingerprints (made of glue, latex, silicone), by burning the fingertip skin (fire, acid, other corrosive material), by using plastic surgery (changing the skin completely, causing change in pattern – portions of skin are removed from a finger and grafted back in different positions, like rotation or “Z” cuts, transplantations of an area from other parts of the body like other fingers, palms, toes, soles). This paper presents a new algorithm for altered fingerprints detection based on fingerprint orientation field reliability. The map of the orientation field reliability has peaks in the singular point locations. These peaks are used to analyze altered fingerprints because, due to alteration, more peaks as singular points appear with lower amplitudes. Keywords: Fingerprints, alteration, image enhancement, reliability, singular points.

Introduction For over 100 years, law enforcement agencies have successfully used fingerprints to identify suspects and victims. Recent advances in automated fingerprint identification technology, coupled with the growing need for reliable person identification, have resulted in an increased use of fingerprints in both government and civilian applications such as border control, employment background checks and secure facility access. The success of fingerprint recognition systems in accurately identifying individuals has prompted some criminals to engage in extreme measures for the purpose of evading identification.

2

Fingerprint alteration is not a new phenomenon. As early as in 1934, John Dillinger, the infamous bank robber and a dangerous criminal, applied acid to his fingertips [1]. Since then, there has been an increase in the reported cases of fingerprint alteration. In 1995, a Criminal was found to have altered his fingerprints by making a ‘Z’ shaped cut into the finger and switching the finger skin the two parts (see Fig. 1). In 2009, a Chinese woman successfully deceived the Japan immigration fingerprint system by performing surgery to swap fingerprints on her left and right hands [3]. Fingerprint alteration has even been performed at a much larger scale involving multiple individuals. Hundreds of asylum seekers have cut, abraded, and burned their fingertips to prevent identification by EURODAC, a European Union fingerprint system for identifying asylum seekers [2]. Additional cases of fingerprint alteration have been compiled in [2]. The primary purpose of fingerprint alteration [1] is to evade identification using techniques that vary from abrading, cutting, and burning fingers to performing plastic surgery. Fingerprint alteration constitutes a serious “attack” against a border control fingerprint identification system since it defeats the very purpose for which the system was deployed in the first place, i.e., to identify individuals on a watch-list. Fingerprint image quality modules used in most fingerprint systems, such as the open source NFIQ (NIST Fingerprint Image Quality) software [4], may be useful in detecting altered fingerprints if the corresponding images are of poor image quality or contain very few minutiae. However, all the altered fingerprint images may not necessarily be of poor quality or contain a small number of minutiae (see Fig. 1). The goal of this work is to introduce the problem of fingerprint alteration and to develop methods to automatically detect and classify altered fingerprints.

Figure 1. A fingerprint altered by switching two parts of a ‘Z’ shaped cut [2].

Literature Review Fingerprint Identification In this section, we first introduce the two fundamental premises of fingerprint identification , which make fingerprints a powerful biometric trait even in the presence of various fingerprint alterations. Then we describe the characteristics of

3

fingerprints that distinguish natural fingerprints from alteredones. Finally, we discuss practical fingerprint identification systems and their vulnerability to altered fingerprints. A. Premises of Fingerprint Identification Permanence and uniqueness are the two fundamental premises that form the basis of friction ridge identification, namely fingerprint and palmprint identification. The friction Fig. 2. Ridge endings (marked with white circles) and ridge bifurcations (marked with black squares). Image is cropped from fingerprint F0134 in NIST SD4 database [4]. ridge skin on human finger, palm, toe and sole consists of two layers: the outer layer, epidermis, and the inner layer, dermis. Both the surface and the bottom of the epidermis contain ridge like formations [5]. The bottom (primary) ridges correspond to the generating (or basal) layer of the epidermis that generates new cells that migrate upwards to the finger surface and slough off. The surface (friction) ridge pattern is a mirror of the bottom ridges, which itself is formed as a result of the buckling process caused by the stress during the growth of fetus at around the fourth month of gestation [6]. Superficial cuts on the surface ridges that do not damage the bottom ridges only temporarily change the surface ridges; after theinjury heals, the surface ridges will grow back to the original pattern. Abrading fingers using rasp only temporarily flattens the friction ridges and they will grow back to the original pattern after some time.

Figure 2 Ridge endings (marked with white circles) and ridge bifurcations (marked with black squares). Image is cropped from fingerprint F0134 in NIST SD4 database [4].

It is generally understood and agreed that friction ridge patterns are not influenced exclusively by genetic factors but also by random physical stresses and tensions that occur during fetal development [7]. These random effects result in the

4

uniqueness property of fingerprints. Even a small portion of friction ridge pattern (e.g., latent fingerprints) contains sufficient detail for establishing one’s identity. In order to evade Identification, fingerprints must be altered to get around these two premises.

B. Characteristics of Fingerprints The friction ridge pattern on a fingertip consists of friction ridges, which are locally parallel and are separated by furrows. The position where a ridge abruptly ends or bifurcates is called a minutia (see Fig. 2). While each fingerprint is unique in detail (such as minutiae and ridge shapes), the overall ridge flow pattern of human fingerprints (and toe prints) is quite similar. Galton classified fingerprints into three basic pattern types: whorl, loop and arch (Fig. 3) [8]. It is believed that such patterns are related to the location and shape of volar pads and the boundary of friction ridges on fingertips (joint crease, and finge nail) [6]. Ridges in whorl and loop fingerprints are separated into three ridge systems: pattern area, distal transverse and proximal transverse systems, by type lines (black lines in Fig. 3) [7]. Delta is the position where three ridge systems meet and core is the innermost position of concentric or loop ridges. Ridge systems in arch fingerprints are not distinguishable.

Figure 3 Three major fingerprint pattern types: (a) whorl, (b) loop, and (c) arch. Three different ridge systems [8] are marked with lines in different colors(red for distal transverse ridges, green for pattern area ridges, and blue for proximal transverse ridges C. Vulnerability of Fingerprint Identification Systems Although the structure of fingerprint patterns can be exploited, to some extent, in order to combat alteration attempts, operational fingerprint identification systems are indeed vulnerable to such attacks. It is very difficult for the state-of-the-art AFIS (Automated Fingerprint Identification Systems) to identify significantly altered fingerprints. Note that it is not necessary to alter the entire friction skin region on the human hand since only a portion of the friction ridge pattern is used in most practical identification systems. Depending on the level of security and the intended application, friction

5

ridge areas recorded and compared by identification systems can vary from the whole hand to a single finger. Furthermore, many non-forensic fingerprint systems use plain (or flat) fingerprint images instead of rolled images. It is also not necessary to completely alter the fingerprints input to automated systems, since their identification accuracy is constrained by image quality, throughput requirements, and database size (false accept rate has to be very low in large-scale systems with millions of enrolled subjects) [9], [10]. Although the accuracy of automated systems in identifying low quality fingerprints can be significantly improved with the help of human operators (as observed in latent identification practice [10]), no specific software is yet available to reconstruct the original pattern of altered fingerprints.

Type Of Altered Fingerprint According to the changes made to the ridge patterns, fingerprint alterations may be categorized into three types: i. Obliteration ii. Distortion iii. Imitation (see Fig. 4). For each type of alteration, its characteristics and possible countermeasures are described.

Figure 4 Three types of altered fingerprints. (a) Obliterated fingerprint (e.g., by burning,) (b) distorted fingerprint (c) imitated fingerprint (simulated by replacing the central region of the original fingerprint with the central region of a different fingerprint

A. Obliteration: Friction ridge patterns on fingertips can be obliterated by abrading , cutting , burning , applying strong chemicals , and transplanting smooth skin . Further, factors

6

such as skin disease (such as leprosy ) and side effects of a cancer drug can also obliterate fingerprints. B. Distortion Friction ridge patterns on fingertips can be turned into unnatural ridge patterns by plastic surgery, in which portions of skin are removed from a finger and grafted back in different positions . Friction skin transplanttation resulting in unnatural ridge patterns also belongs to this category. This type of fingerprint alteration has been increasingly observed in border control applications. Therefore, it is imperative to upgrade current fingerprint quality control software to detect this type of altered fingerprints. Once detected, the following actions may be taken to assist the automated fingerprint matcher: (i) identify unaffected regions of the fingerprint and manually mark features (i.e., the minutiae) in these regions and (ii) reconstruct the original fingerprint as done by the latent examiner in the ‘Z’ cut case . C. Imitation Here, a surgical procedure is performed in such a way that the altered fingerprints appear as a natural fingerprint ridge pattern. Such surgeries may involve the transplantation of a large-area friction skin from other parts of the body, such as fingers, palms, toes, and soles (see Fig. 1a and simulation in Fig. 5), or even cutting and mosaicking multiple small portions of friction skin (see simulation in Fig. 6).

Figure 5 Simulation of large-area transplantation between two fingerprints: (a) Original fingerprint (b) altered fingerprint by transplanting central area

Transplanted fingerprints can successfully evade existing fingerprint quality control software. If the surgical scars due to the transplantation are small, it can even deceive inexpe- rienced human operators. As long as the transplanted area is large, matching altered fingerprints to the original (unaltered) fingerprints is not likely to succeed. Plain images captured by fingerprint scanners used in most border control

7

applications may not be able to reveal the surgical scars in large-area transplantation. But the large-area transplantation has the risk of being matched to the donor print (if the donor print that is contained in the database is also searched). Further, reconstructing the original fingerprint is not difficult since transplantation is generally performed using friction skin of the same person as in the Marc George’s case. Small area transplantation is probably a more complicated surgery.

\

Pro-

posed

Figure 6 Simulation of small-area transplantation within a finger. (a) Original fingerprint and (b) altered fingerprint. Simulation is performed by exchanging and rotating circular regions (marked with the same number) to match the local ridge orientation or just rotating circular regions(marked with number 2) by 180 degrees Work The success of automated fingerprint identification systems has prompted some individuals to take extreme measures to evade identification by altering their fingerprints. The problem of fingerprint alteration or obfuscation is very different from that of fingerprint spoofing where an individual uses a fake fingerprint in order to adopt the identity of another individual. While the problem of spoofing has received increased attention in the literature, the problem of obfuscation has not been discussed in the biometric literature in spite of numerous documented cases of fingerprint alteration to evade identification. The lack of public databases containing altered fingerprints has further stymied research on this topic. While obfuscation may be encountered in biometric systems adopting other types of modalities (such as face and iris), this problem is especially significant in the case of fingerprints due to the widespread deployment of fingerprint systems in both government and civilian application and the ease with which these “attacks” can be launched. We have introduced the problem of finger- print obfuscation and discussed a categorization scheme to characterize the various types of altered fingerprints that have been observed. It is desirable to develop a method that can automatically detect altered fingerprints. Available fingerprint quality control software modules have very limited capability in distinguishing altered fingerprints from natural fingerprints. Here

8

we proposed an algorithm to automatically detect altered (distorted) fingerprints and classify according to its type. The underlying idea is that altered fingerprints often show unusual ridge patterns.

Methodology The NFIQ algorithm is not suitable for detecting altered fingerprints, especially the distortion and imitation types. In fact, the distorted and imitated fingerprints are very hard to detect for any fingerprint image quality assessment algorithm that is based on analyzing local image quality[11]. In this section, we consider the problem of automatic detection of alterations and classification based on analyzing singular point reliability map of orientation field. The flowchart of the proposed alteration type detector is given in Fig. 7. Flow Chart for proposed Algorithm A set of features is first extracted from the ridge orientation field of an input fingerprint and then a fuzzy classifier is used to classify it into natural or altered fingerprint and its alteration type. Fingerprint

Preprocessing on Image

Orientation field Estimation

Orientation field Reliability map

Detection of SP’s

N

Fuzzy Classification Y

Y Altered

Fuzzy Classification

Obliteration

Distortion

Imitation

Figure 7. Flow Chart for proposed Algorithm

9

Fingerprint Database Due to lack of altered fingerprint data, I use constructed database synthetically form the original fingerprints. I collect more than hundred original fingerprints of my friends to analyze the proposed system. After collecting the original fingerprint, I altered those fingerprint by the three different techniques of fingerprint alteration named obliteration, distortion and imitation respectively. For analyzing the system we use 49 images for all type of alteration. Preprocessing on Fingerprint The aim of the image pre-processing stage is to increase both the accuracy and the interpretability of the digital data during the image processing stage. The preprocessing takes place prior to any principal component analysis. The main steps for pre-processing are enhancement, binarization, distance transform and segmentation. First, a pre-processing stage of image enhancement is performed in order to improve the contrast between ridges and valleys and reduce noises in the fingerprint images. Two methods are adopted for image enhancement stage: the first one is based on Histogram Equalization and the next one on Fourier Transform. A local adaptive binarization method is performed to binarize the fingerprint image. The binary image obtained after image enhancement is transformed into gray scale image using the euclidean distance transform. Fingerprint segmentation is implemented, to decide which part of the image belongs to the foreground and which part belongs to the background. The singular points will be located more accurately if the localization operates only on the foreground of the fingerprint image. Image Enhancement The fingerprint input image is enhanced in the spatial domain by applying the Histogram Equalization technique, for a better distribution of the pixel values.

10

Considering the real altered fingerprint from Fig. 8(a), the result image is represented in Fig. 8(b)

(a)

(b)

(d)

(e)

(c)

(f)

Fig. 8 Pre-processing steps: a). Real altered fingerprint image; b). Image obtain after Histogram Equalization; c).Image obtained after FFT; d). Binary image; e). Distance transform image; f). Region of interest.

Histogram Equalization The histogram of a digital image with gray levels in the range [0, L-1] is a discrete function: (1) where represents the k ’th gray level, is the number of pixels in the image with that gray level, n represents the total number of pixels in the image, k=1,1,….L,L=256.

Fast Fourier transform In the frequency domain, the fingerprint image is enhanced based on the Fast Fourier Transform (FFT), Fig. 8(c). The image is divided into 32 *32 processing blocks and the Fourier Transform is performed according to:

11

For u= 0,1,……31 and v= 0,1,……..31.. In order to enhance a specific block by its dominant frequencies, the FFT of the block is multiplied by its magnitude a set of times, where the magnitude of the original FFT is:

The enhanced block is obtained according to:

Where
 
  is computed using:

for u= 0,1,……31 and v= 0,1,……..31.. The experimentally constant k is set to 0.45 (a higher k improves the appearance of the ridges, filling up small holes in ridges; a very high k may introduce false joining of ridges). Binarisation Most minutiae extraction algorithms operate on basically binary images where there are only two levels of interest: the black pixels represent ridges, and the white pixels represent valleys. Binarisation [1] converts a greylevel image into a binary image. This helps in improving the contrast between the ridges and valleys in a fingerprint image, and consequently facilitates the extraction of minutiae. One very useful property of the Gabor filter is that it contains a DC component of zero, which indicates that the resulting altered image has a zero mean pixel value. Hence, binarisation of the image can be done by using a global threshold of zero. Binarisation involves examining the grey-level value of every pixel in the enhanced image, and, if the grey-level value is greater than the pre denned global threshold, then the pixel value is set to value one; else, it is set to zero. The out-

12

come of binarisation is a binary image which contains two levels of information, the background valleys and the foreground ridges. Distance Transform of Binary image The distance transform of the binary image, Fig. 1(d), is defined as the distance from every black pixel (belonging to the fingerprint ridges) to the nearest white pixel. The euclidean distance between two pixels (i1, j1 ) and (i2, j2) is defined as :

In order to mark the foreground and background images, a block wise ( 8*8 pixels) binary image is created based on the distance transform image, represented in Fig. 1(e). A block of 8* 8 pixels is set as foreground if at least 80% of its pixels have values smaller than a threshold (set to 10.), Fig. 1(f). For each block in the foreground, the orientation field is estimated applying a gradient-based method, based on the distance transform image. Ridge Orientation Estimation The orientation field image is a Level 1 feature that represents the angle θi,j shown in Fig. 9, that the finger print ridges form with the horizontal axis, crossing through an arbitrary small neighborhood centered at (i,j)

Fig-9 Ridge ending and ridge bifurcation orientation angle This is a gradient based method (Hong et al., 1998). A single orientation is assigned for each non-overlapping block of size w ×w (16 ×16 pixels) that corresponds to the dominant orientation of the block. The gradient G at the point (i,j) is computed as a 2 dimensional vector with the components Gx(i,j) and G y(i,j) the

13

horizontal and the vertical gradients, with respect to the x and y directions. Simple gradient operators are used such as a Sobel mask ( 3×3 ). The local orientation of each block centered at pixel (i,j) is estimated using the following equations: Where θ(i,j) is the least square estimate of the local ridge orientation at the

block centered at pixel (i,j). In order to adjust the incorrect local ridge orientation, due to the presence of noise, a low-pass filtering can be used, since local ridge orientation varies slowly in a local neighborhood where singular points appear. The orientation image needs to be converted into a continuous vector field, as follows:

Where Φx(i,j) and Φy(i,j) are components of the vector field. With the resulting vector field, the low-pass filtering can be performed as follows:

14

Where W is a 2-dimensional low-pass filter with unit integral and wΦ × wΦ specifies the size of the filter (default size 5x5 ). After the smoothing operation is performed at the block level, the local ridge orientation at ( i,j) is computed using:

Where O(i,j) represents the orientation image. Singular Point Detection Singular point detection is the most challenging task; it is an important process for fingerprint image alignment, fingerprint classification and fingerprint matching. In the following subsections, we propose orientation reliability and singular point position methods. Orientation field Reliability Map The fingerprint image is made up of pattern of ridges and valleys; they are the replica of the human fingertips. The fingerprint image represents a system of oriented texture and has very rich structural information within the image. This flow-like pattern forms an orientation field extracted from the style of valleys and ridges. In the large part of fingerprint topologies, the orientation field is quite smooth, while in some areas, the orientation appears in a discontinuous manner. These regions are called singularity or singular points, including core and delta and are defined as the centers of those areas. In addition, the reference point is defined here as the point with maximum curvature on the convex ridge.

15

The reliability of the orientation filed describes the consistency of the local orientations in a neighborhood along the dominant orientation is used to locate the unique reference point constantly for all types of fingerprints. The reliability can be also computed using the coherence as proposed by (Kaas and Witkin, 1987) and (Bazen and Gerez ,2002). The implementation is elaborated on in the following: 1.

The orientation image is hardly ever computed at full-resolution. Instead each non-overlapping block of size W × W of the image is assigned a single orientation that correspond to the most apparent or dominant orientation of the block. In this proposed method, W is set equal to sixteen.

2.

The horizontal and vertical gradients Gx(x, y) and Gy(x, y) at each pixel (x,y) respectively are computed using simple gradient operators such as a Sobel mask (Gonzalez and Woods; 2008). The mask is set to 3 X 3.

3.

Compute the ridge orientation of each pixel (x,y) by averaging the squared gradients within a W × W window centered at [xi,yj] as follows (Ratha et al., 1995):





     

         

          

4.

         

  Because of noise, corrupted ridge, valley structures and low gray value contrast, a low-pass filter can be used to adjust the erroneous local ridge orientation. However, to perform the low-pass filtering, the orientation image needs to be converted into a continuous vector field as follows:

16

And



 !"#   $

  #%  & where and  are the x and y components of the vector field, respectively. With the resulting vector field, the Gaussian low-pass filter can be applied as follows:

5.





 

 ' 



 '   '(

  

 ' 

 

 '   ',

+* )*   +* )*

Since the singular point has the maximum curvature. It can be located by measuring the strength of the peak using the following:

-% ./0123/4   ((  5

  6

  7     8 9 -: ./0123/4    5

 -% ./0123/4 ; -% ./012/4  /=/3      -: ./0123/4 

17

Figure-10 – Screen shots of implimatation of proposed work

18

Result Analysis The natural fingerprints are compared with the altered fingerprints, by observing the values of the reliability singular points. In this dissertation, the singular point is defined as the point with maximum curvature on the convex ridge. For natural fingerprints, the reliability orientation image generally has one sharp point, while in altered fingerprints more point are detected with smaller values. Starting from this observation, the altered fingerprint analysis can be done using the density and the count of the singular points. The proposed algorithm for fingerprint analysis based on the estimation of orientation field and the computation of the reliability was tested with real altered fingerprint and simulated altered fingerprint obtained from natural fingerprint images by using synthetic method due to unavailability of altered fingerprint database. The real altered fingerprint by distortion shaped is shown in Fig 11(a). Experimental results show that multiple singular points are detected shown in fig.11 (b)

Figure-11 (a) Singular point estimation using reliability map (original)

Figure-11 (b) Singular point estimation using reliability map (Altered by obliteration)

19

Figure-11 (b) Singular point estimation using reliability map (Altered by Distortion)

.

Figure-11 (b) Singular point estimation using reliability map (Altered by Imitation) In (Feng et al., 2009), it was pointed out that using a feature vector termed as curvature histogram, extracted from the continuous vector field, classification, leads to results showing that 92% altered fingerprints can be correctly detected. In this paper, singular points are detected with high amplitudes for natural fingerprints, and relatively small amplitudes for altered fingerprints, that can be used for classifying algorithms, in order to give a completely comparative study. Testing the efficiency of the orientation field reliability, the experimental results indicate that the reliability R has strong information that can be used for future research. The singular value decomposition of R leads to obtaining essential features for discrimination and has good stability. The results obtained in altered

20

fingerprints analysis using orientation field reliability are persuasive and could be employed for automatic detection of biometric obfuscation. The orientation field reliability map has peaks in the singular point locations and these peaks are used to analyze altered fingerprints. Due to alteration, more numbers of singular points appear with lower amplitudes. The experimental results demonstrate that the proposed algorithm can provide important information in order to automatically detect altered fingerprints.

Figure-12 Fuzzy test on database (Original and Altered fingerprint)

Figure-13 the ROC curves of the proposed algorithm and the NFIQ algorithm for each type of altered fingerprints. The ROC curve of the NFIQ criterion is shown as a set of points because its output can only take one of the five quality levels.

21

Table 1. Comparison of the proposed method with other methods

Method Kawagoe and Tojo (1984)

False SP (%) 7.5

Missed SP(%) 10.4

Zheng et al. (2006)

4.9

Rahimi et al. (2004)

12.5

25

Wang and Dai (2006)

3.2

8.4

Chikkerur and Ratha (2005)

4.4

8.3

Proposed method

2

1.5

7.8

. Table 1 shows the comparison between the proposed method and the methods proposed by Kawagoe and Tojo (1984) based on the Poincare index, which is derived from continuous curves, Zheng et al. (2006) using the combination of curvature and orientation field of fingerprint images based on Poincaré index, Rahimi et al. (2004) using the directional image and absolute correlation function, Wang and Dai (2006) used Gaussian-Hermite moments (GHMs) behavior with Poincare index and Chikkerur and Ratha (2005) based on the complex filtering principles; it indicates clearly that the proposed method is more accurate than other methods. The maximum difference between the compared methods for false detection is 10.5 and 8.9% for missing detection. The minimum difference between the compared methods for false detection is 1.2 and 6.3% for missing detection.

Conclusion The proposed algorithm will be tested using altered fingerprints synthesized in the way typically observed in operational cases with good performance. This dissertation proposes a novel method to consistently and precisely locate the singular points (core and delta) in fingerprint images. The method applied is based on the enhanced fingerprint image orientation reliability. In addition, an enhancement for the fingerprint image is applied using the Short Time Fourier Transform analysis (STFT). The experimental results demonstrate that the proposed algorithm is more accurate than the methods based on the Poincare index, which is derived from continuous curves, combination of curvature and orientation field of fingerprint images based on Poincare Index, the directional image and absolute correlation function and Gaussian–Hermite moments (GHMs) behavior with Poincare Index. The result for the proposed

22

method is 2% locating the singular points with spurious, 1.5% missing locating the singular points. This method also locates a secondary core and delta if it exists. Our future work will focus on improvements in locating the secondary singular points of fingerprint images.

Future Scope And Limitations The current altered fingerprint detection algorithm can be improved along the following directions: 1. Determine the alteration type efficiently and automatically so that appropriate countermeasures can be taken. 2. Reconstruct altered fingerprints. For some types of fingerprints where the ridge patterns are damaged locally or the ridge structure is still present on the finger but possibly at a different location, reconstruction is indeed possible. 3. Match altered fingerprints to their unaltered mates. A matcher specialized for altered fingerprints can be developed to link them to unaltered mates in the database utilizing whatever information is available in the altered fingerprints. In this dissertation we use the synthetic fingerprints due to unavailability of public databases, but this method will give a way for researcher to do in the fields of altered fingerprint recognition.

References [1] H. Cummins, “Attempts to Alter and Obliterate Fingerprints,” Journal of American Institute of Criminal Law and Criminology, vol. 25, pp. 982–991, 1935. [2]K. Singh, “Altered Fingerprints,” 2008. http://www.interpol.int/Public/Forensic/fingerprints/research/alteredfingerprints.pdf [3] Surgically Altered Fingerprints Help Woman Evade Immigration, Dec. 2009. http://abcnews.go.com [4] NIST Special Database 4, NIST 8-Bit Gray Scale Images of Fingerprint Image Groups (FIGS), http://www.nist.gov/srd/nistsd4.htm. [5] D. R. Ashbaugh, Quantitative-Qualitative Friction Ridge Analysis: An Introduction to Basic and Advanced Ridgeology. CRC Press, 1999. [6] M. Kcken and A. C. Newell, “Fingerprint Formation,” Journal of Theoretical Biology, vol. 235, no. 1, pp. 71 – 83, 2005. [7 H. Cummins and M. Midlo, Finger Prints, Palms and Soles: An Introduction to Dermatoglyphics. New York: Dover Publications, 1961. [8] F. Galton, Finger Prints (reprint). New York: Da Capo Press, 1965. [9] C. L. Wilson, M. D. Garris, and C. I. Watson, “Matching Performance for the US-VISIT IDENT System Using Flat Fingerprints,” NISTIR 7110, May 2004, ftp://sequoyah.nist.gov/pub/nist internal reports/ir7110.pdf.

23 [10] L. M.Wein and M. Baveja, “Using Fingerprint Image Quality to Improve the Identification Performance of the U.S. Visitor and Immigrant Status Indicator Technology Program,” Proc. National Academy of Sciences of the U.S.A., vol. 102, no. 21, pp. 7772– 7775, 2005 [11] Anil K Jain, Soweon Yoon, Jianjiang Feng ,”Altered Fingerprints:Analysis and De tection”,IEEE transaction on pattern analysis and machine intelligence , Vol.34, 2012