Data Compression of Fingerprint Minutiae VISHAL SHRIVASTAVA Department of Electronics and Communication, Shri Ram Institute of Technology, Near ITI, Madhotal, Jabalpur M.P., 482002, India Vishal11shrivastava@gmail.com http://www.sritedu.org SUMIT SHARMA Department of Electronics and Communication, Shri Ram Group of Institutes, Near ITI, Madhotal, Jabalpur, M.P., 482002, India sharma.sumit3@gmail.com http://www.sritedu.org Abstract: Biometric techniques have usual advantages over conventional personal identification technique. Among various commercially available biometric techniques such as face, fingerprint, Iris etc., fingerprint-based techniques are the most accepted recognition system. Fingerprints are trace or impression of patterns created by friction ridges of the skin in the fingers and thumbs. Steganography usually used in smart card is a safe technique for authenticating a person. In steganography, biometric characteristic similar to fingerprint is hidden in an picture. As the quantity of information that can be stored by means of steganography is extremely restricted, compression mechanisms are essential in order to achieve reasonably little errors when finally checking fingerprints against the encoded templates. To decrease the volume of smart card, compression can be applied to fingerprint template in order to make it take up fewer space. This paper is presenting the minutiae based inexpensive fingerprint compression technique. In minutiae based systems, the discontinuities in the regular ridge structure of fingerprint images are acknowledged in feature extraction step. During matching, a similarity value between the features extracted from the template and the enter fingerprint images is calculated. This similarity value is used to appear at an accept / reject decision. We present a new approach based on delta compression for robustly compressing the fingerprint templates. Keywords: Fingerprint templates, Steganography, Fingerprint Minutiae, Delta Compression technique. 1. Introduction Smart card uses special security ways for authentication. We can use picture of face, fingerprint etc., for security purpose, which is helpful when a person show his smart card for authentication. To, decrease the volume of smart card compression is required. A different technique as watermarking/steganography is used to store this biometric information in smart card [10]. As well as, we can also add little other information. As Heikki Ailisto has join the body weight and fat measurement with fingerprint biometrics [4]. To compress the data, different compression algorithms are existing. According to type of data, we can select any one of them. We compress the fingerprint or face picture with definite compression ratios. When we conduct a fingerprint or face recognition procedure, matching the compressed templates against uncompressed original image data, severe distortion may happen. And, it might result fake rejection or non-match. We relate rate distortion performance as measured in PSNR to the matching scores as obtained by recognition system. To store the picture like fingerprint or face we can use JPEG compression technology [14]. But, PSNR suggests JPEG to deliver worse recognition result [13].In this paper, compression of minutiae based fingerprint template is proposed which can be used in smart card. 2. Biometric & Fingerprint Minutiae Presently time biometric is one of the best method to recognize a person. ISO/IEC JTC 1/SC 37 has developed different standard for biometric data and their interfacing [2]. In biometric characteristic, we capture the physical or behavioral characteristic of a person for identification [15]. The processes of biometric recognition system have four steps [10]: Acquisition, Representation, Feature Extraction, and Matching. In the first step, biometric characteristic as fingerprint picture is access by the scanner. In biometric recognition system any type of physiological or behavioral characteristic like fingerprint, iris, and signature dynamics can be used. In the representation stage, the object is to find invariant information of fingerprints. In the third step, we try to get any type of irregularities as minutiae in fingerprint. During, matching, a similarity Vishal Shrivastava et al. / International Journal of Engineering Science and Technology (IJEST)ISSN : 0975-5462Vol. 4 No.02 February 2012401
between input image of biometric characteristic and features extracted and stored as a template is measured. Biometric recognition can be classified as identification or authentication [9]. Identification occurs when biometric characteristic of an individual is matched with all the available templates saved in database. So, here one to many search Is done. But, in the case of authentication, an individual claim for his identity by entering a password or by showing his card. So, in the case of authentication, individual characteristic is matched only with enrolled characteristic. So, one to one searching is done, hence process is much faster. Fig. 1. Fingerprint Minutiae. Fingerprint is one of the most popular biometric recognition characteristic. ISO/IEC has published the standard ISO/IEC 19794-2 which specifies a concept and data formats for representation of fingerprints by means of the basic notion of minutiae [2]. Most of the fingerprint identification /authentication algorithm is based on minutiae present in the fingerprint. Minutiae are natural irregularities present in the fingerprint like ridge bifurcation, ridge ending, delta core, etc.. Ending is the point at which a ridge stops. Bifurcation is the point at which one ridge divides into two [15]. As well as, there are many minutiae present in fingerprint as shown in figure 1. 2.1. Minutiae Identification To identify the minutiae we capture input image of a person using a sensor. Then, this image is digitised and converted into grey shade. Now, by identifying the no. of neighbour we can find the minutiae. In the fingerprint bifurcation, there are three neighbours with grey shade. In the case of ridge end, there is only single neighbour present as shown in the figure 2 [7]. Ridge ending and ridge bifurcation are the most popular minutiae used for fingerprint identification. Fig. 1. (a) Bifurcation (b) Ridge end identification Vishal Shrivastava et al. / International Journal of Engineering Science and Technology (IJEST)ISSN : 0975-5462Vol. 4 No.02 February 2012402
2.2 Cost-effective minutiae based fingerprint identification In most commercially available system, it is the hardware element that renders the systems highly expensive [8]. In cost-effective fingerprint identification we use fewer qualitative sensors. So, to remove this type of error we can eliminate the fourth attribute i.e. type of minutiae. In this way, we can use poor quality hardware and sensor as well as save the memory space. 3. Delta code Sometimes, there may appear patterns of members that are practically unpredictable, but with neighboring terms close to each other, such as readings of temperature. What one could do is to trace the first value, and from then on record the difference to the next. For example: {23, 27, 25, 24, 21, 19, 22, 22, 24, 27, 26} = {23, +4, –2, –1, –3, –2, +3, +0, +2, +3, –1} Since the increments are from zero to three, we only require two bits to store them [6]. Note, however that the increments can be either positive or negative. This can be handled by the concept of negative binary numbers, where all integers – positive, zero, and negative can be represented as whole numbers. 4. Steganography Steganography is a method to hide the information in any picture [10]. This information can be extract by a secret key. In case of fingerprint recognition, fingerprint minutiae can be stored in any image. This is one of the secure technique which is usually used in smart card. In this technique, data like fingerprint minutiae are hidden. Hence, it is more secure, because any imposter can not understand that fingerprint is hidden in the image. Here, same key is used to hide the fingerprint minutiae and to decode those minutiae at the time of verification. These decoded minutiae are matched with original input fingerprint of the person for authentication. 5. Compression of Fingerprint Minutiae Anil K. Jain has presented the method of steganography to hide the fingerprint minutiae in a host picture [10]. This job is extension of that algorithm to save memory space. The bulk of smart card must be small. So, there is limited memory space to save the fingerprint minutiae. Hence compression techniques are required. Generally, 24 to 28 minutiae are extract by the fingerprint. Figure 4, shows the host picture used to hide the fingerprint, the fingerprint template and table which point out the position of minutiae in the fingerprint templates. Usually 9-bits are used in each field (x-coordinate, y-coordinate and rotation angle) to show the position of minutiae in between 0 to 511. Fig. 4. (a) Input Fingerprintimage (b) Overlaid Minutiae image (c) Sample host image To decrease the memory amount, we have used delta codes for x-coordinate. In delta code, we store the first value and from then on record the difference to the next. Before, uses of delta code, minutiae are arranged in ascending order according to their x-field value. Its benefit is that, we can save one bit for each code, which is set to give you an idea about sign of the difference. With delta compression, for x-coordinate 20% memory space can be saved. Also, to store the minutiae point, we do not utilize fourth field i.e. type of minutiae to reduce the memory size. And, in this way, we can overcome the chances of error. If, a ridge ending look like a bifurcation or vice-versa. Then, it will not generate any error for authentication. Vishal Shrivastava et al. / International Journal of Engineering Science and Technology (IJEST)ISSN : 0975-5462Vol. 4 No.02 February 2012403
Table 1. Minutiae points in the fingerprint. S. No. X - field Delta codes for X-field Y - field Rotation angle 1 60 60 92 356 2 76 16 216 242 3 77 1 197 58 4 88 11 85 144 5 98 10 69 332 6 121 23 195 255 7 136 15 82 292 8 136 0 229 248 9 170 34 90 262 10 172 2 169 270 11 178 6 46 274 12 184 6 85 82 13 192 8 146 281 14 196 4 198 270 15 201 5 89 52 16 212 11 233 255 17 216 4 220 262 18 228 12 125 321 19 234 6 79 8 20 234 0 147 298 21 236 4 175 295 22 239 3 190 274 23 240 1 167 112 24 251 11 222 270 25 259 8 68 356 In this example, we have used 5-bits to show x-coordinate position of first minutiae. And for remaining position we have used 4 bits. With the help of four bits we can illustrate the difference from 0 to 15 which is sufficient to represent the position of next minutiae is 675 bits. And with delta code it has reduce up to 555 bits. 6. Conclusion By delta compression, we can hide fingerprint minutiae information, with less number of bits in steganography. Therefore, we can decrease the memory amount and that's why size of the smart card. By the utilization of the proposed technique more than 20% data can be compressed. References [1] Mohammad Al-laham, Ibrahiem M. M. El Emary: Comparative Study between Various Algorithms of Data Compression Techniques. IJCSNS, Vol. 7 No. 4, April 2007. [2] International Standard Developed by ISO/IEC JTC 1/SC 37 – Biometrics. Revised, 2007. [3] Andreas Chwatal, Gunther Raidl, Olive Dietzel: Compressing Fingerprint Templates by Solving an Extended Minimum Label Spanning Tree Problem (MIC 2007). [4] Heikki Ailisto, Elena Vildjiounaite, Mikko Lindholm, Satu-Marja Makela, Johnnes Peltola: Soft Biometrics – Combining Body Weight and Fat Measurements with Fingerprint Biometrics. Pattern Recognition Letters 27, 325 – 334 (2006). [5] Zhongqing Su, Lin Ye: Digital Damage Fingerprints (DDF) and Its Application in Quantitative Damage Identification. Composite Structures 67 197 – 204 (2005) [6] Lucas Garron: Data Compression. August 4 (2005). [7] Salil Prabhakar, Anil K. Jain, Sharath Pankanti: Learning Fingerprint Minutiae Location and Type. Pattern recognition 36 (2003) 1847 – 1857. [8] A. J. Wills, L. Myers: A Cost-effective Fingerprint Recognition System for Use with Low-quality Prints and Damaged Fingertips. Pattern recognition 34 (2001) 255 – 270. [9] By Erik Bowman: Everything You Need to Know About Biometrics. Identix Corporation January 2000. [10] Anil K. Jain and Umut Uldag: Hiding Fingerprint Minutiae in Images.
between input image of biometric characteristic and features extracted and stored as a template is measured. Biometric recognition can be classified as identification or authentication [9]. Identification occurs when biometric characteristic of an individual is matched with all the available templates saved in database. So, here one to many search Is done. But, in the case of authentication, an individual claim for his identity by entering a password or by showing his card. So, in the case of authentication, individual characteristic is matched only with enrolled characteristic. So, one to one searching is done, hence process is much faster. Fig. 1. Fingerprint Minutiae. Fingerprint is one of the most popular biometric recognition characteristic. ISO/IEC has published the standard ISO/IEC 19794-2 which specifies a concept and data formats for representation of fingerprints by means of the basic notion of minutiae [2]. Most of the fingerprint identification /authentication algorithm is based on minutiae present in the fingerprint. Minutiae are natural irregularities present in the fingerprint like ridge bifurcation, ridge ending, delta core, etc.. Ending is the point at which a ridge stops. Bifurcation is the point at which one ridge divides into two [15]. As well as, there are many minutiae present in fingerprint as shown in figure 1. 2.1. Minutiae Identification To identify the minutiae we capture input image of a person using a sensor. Then, this image is digitised and converted into grey shade. Now, by identifying the no. of neighbour we can find the minutiae. In the fingerprint bifurcation, there are three neighbours with grey shade. In the case of ridge end, there is only single neighbour present as shown in the figure 2 [7]. Ridge ending and ridge bifurcation are the most popular minutiae used for fingerprint identification. Fig. 1. (a) Bifurcation (b) Ridge end identification Vishal Shrivastava et al. / International Journal of Engineering Science and Technology (IJEST)ISSN : 0975-5462Vol. 4 No.02 February 2012402
2.2 Cost-effective minutiae based fingerprint identification In most commercially available system, it is the hardware element that renders the systems highly expensive [8]. In cost-effective fingerprint identification we use fewer qualitative sensors. So, to remove this type of error we can eliminate the fourth attribute i.e. type of minutiae. In this way, we can use poor quality hardware and sensor as well as save the memory space. 3. Delta code Sometimes, there may appear patterns of members that are practically unpredictable, but with neighboring terms close to each other, such as readings of temperature. What one could do is to trace the first value, and from then on record the difference to the next. For example: {23, 27, 25, 24, 21, 19, 22, 22, 24, 27, 26} = {23, +4, –2, –1, –3, –2, +3, +0, +2, +3, –1} Since the increments are from zero to three, we only require two bits to store them [6]. Note, however that the increments can be either positive or negative. This can be handled by the concept of negative binary numbers, where all integers – positive, zero, and negative can be represented as whole numbers. 4. Steganography Steganography is a method to hide the information in any picture [10]. This information can be extract by a secret key. In case of fingerprint recognition, fingerprint minutiae can be stored in any image. This is one of the secure technique which is usually used in smart card. In this technique, data like fingerprint minutiae are hidden. Hence, it is more secure, because any imposter can not understand that fingerprint is hidden in the image. Here, same key is used to hide the fingerprint minutiae and to decode those minutiae at the time of verification. These decoded minutiae are matched with original input fingerprint of the person for authentication. 5. Compression of Fingerprint Minutiae Anil K. Jain has presented the method of steganography to hide the fingerprint minutiae in a host picture [10]. This job is extension of that algorithm to save memory space. The bulk of smart card must be small. So, there is limited memory space to save the fingerprint minutiae. Hence compression techniques are required. Generally, 24 to 28 minutiae are extract by the fingerprint. Figure 4, shows the host picture used to hide the fingerprint, the fingerprint template and table which point out the position of minutiae in the fingerprint templates. Usually 9-bits are used in each field (x-coordinate, y-coordinate and rotation angle) to show the position of minutiae in between 0 to 511. Fig. 4. (a) Input Fingerprintimage (b) Overlaid Minutiae image (c) Sample host image To decrease the memory amount, we have used delta codes for x-coordinate. In delta code, we store the first value and from then on record the difference to the next. Before, uses of delta code, minutiae are arranged in ascending order according to their x-field value. Its benefit is that, we can save one bit for each code, which is set to give you an idea about sign of the difference. With delta compression, for x-coordinate 20% memory space can be saved. Also, to store the minutiae point, we do not utilize fourth field i.e. type of minutiae to reduce the memory size. And, in this way, we can overcome the chances of error. If, a ridge ending look like a bifurcation or vice-versa. Then, it will not generate any error for authentication. Vishal Shrivastava et al. / International Journal of Engineering Science and Technology (IJEST)ISSN : 0975-5462Vol. 4 No.02 February 2012403
Table 1. Minutiae points in the fingerprint. S. No. X - field Delta codes for X-field Y - field Rotation angle 1 60 60 92 356 2 76 16 216 242 3 77 1 197 58 4 88 11 85 144 5 98 10 69 332 6 121 23 195 255 7 136 15 82 292 8 136 0 229 248 9 170 34 90 262 10 172 2 169 270 11 178 6 46 274 12 184 6 85 82 13 192 8 146 281 14 196 4 198 270 15 201 5 89 52 16 212 11 233 255 17 216 4 220 262 18 228 12 125 321 19 234 6 79 8 20 234 0 147 298 21 236 4 175 295 22 239 3 190 274 23 240 1 167 112 24 251 11 222 270 25 259 8 68 356 In this example, we have used 5-bits to show x-coordinate position of first minutiae. And for remaining position we have used 4 bits. With the help of four bits we can illustrate the difference from 0 to 15 which is sufficient to represent the position of next minutiae is 675 bits. And with delta code it has reduce up to 555 bits. 6. Conclusion By delta compression, we can hide fingerprint minutiae information, with less number of bits in steganography. Therefore, we can decrease the memory amount and that's why size of the smart card. By the utilization of the proposed technique more than 20% data can be compressed. References [1] Mohammad Al-laham, Ibrahiem M. M. El Emary: Comparative Study between Various Algorithms of Data Compression Techniques. IJCSNS, Vol. 7 No. 4, April 2007. [2] International Standard Developed by ISO/IEC JTC 1/SC 37 – Biometrics. Revised, 2007. [3] Andreas Chwatal, Gunther Raidl, Olive Dietzel: Compressing Fingerprint Templates by Solving an Extended Minimum Label Spanning Tree Problem (MIC 2007). [4] Heikki Ailisto, Elena Vildjiounaite, Mikko Lindholm, Satu-Marja Makela, Johnnes Peltola: Soft Biometrics – Combining Body Weight and Fat Measurements with Fingerprint Biometrics. Pattern Recognition Letters 27, 325 – 334 (2006). [5] Zhongqing Su, Lin Ye: Digital Damage Fingerprints (DDF) and Its Application in Quantitative Damage Identification. Composite Structures 67 197 – 204 (2005) [6] Lucas Garron: Data Compression. August 4 (2005). [7] Salil Prabhakar, Anil K. Jain, Sharath Pankanti: Learning Fingerprint Minutiae Location and Type. Pattern recognition 36 (2003) 1847 – 1857. [8] A. J. Wills, L. Myers: A Cost-effective Fingerprint Recognition System for Use with Low-quality Prints and Damaged Fingertips. Pattern recognition 34 (2001) 255 – 270. [9] By Erik Bowman: Everything You Need to Know About Biometrics. Identix Corporation January 2000. [10] Anil K. Jain and Umut Uldag: Hiding Fingerprint Minutiae in Images.
No comments:
Post a Comment