International Journal of Scientific & Engineering Research, Volume 4, Issue 7, July-2013 2068

ISSN 2229-5518

A Comparative study of Palmprint Feature Extraction

Mrs. Kasturika B. Ray

Dean (Academics)

Eklavya College of Engineering Technology and Science

Kusumati , Bhubaneswar , Odisha , India Institute of Technical Education & Research. Siksha ‘O’ Anusandhan University Bhubaneswar (Phd contd.)

E-mail – kasturikabray@rediffmail.com

Abstract— The different kind of characteristics of human beings such as fingerprints, palm print , facial features , iris pattern , retina and hand geometry are the personal identification and verification computing by Biometric. Such type of Biometric verification and authentication approach using the biological features inherent in each individual. This kind of Biometric identification is a technology which describes the general procedure for verification using feature extraction. The palm print features are introduced some interesting points and textural features by Biometrics. It is very suitable in many network based applications. Now it is a widespread method of identification and authentication uses several image processing techniques. In this paper we present a comparative study and analysis of some palm print feature extraction and identification methods.

Keywords—Personal authentication, Biometric computing, Interesting points, feature extraction, image matching, palm print features.

Introduction

Biometric features have been widely used in many personal authentication applications like universality, Distinctiveness, Permanence, Collectability, Performance, Acceptability and Circumvention [1]. Biometric features are the features extracted from human biological organs
or behaviour for personal identification by Jain et al in 1999 [2]. There has been a high demand for personal identification and verification for security reasons and different kind of characteristics of human beings such as fingerprints, palm print, facial features, iris pattern, retina and hand geometry are the personal identification and verification computing by Biometric [3]. Some researchers have investigated the problem of logo recognition, by applying algebraic and differential invariants [4] , positive and negative shape feature [5], Zernike moment [6], String- Matching technique [7], template matching [8], combined measure [9], Zhang proposed a texture based feature extraction method to obtain the global attributes of a palm, besides a dynamic selection scheme was also designed to ensure the palm print samples to be correctly and electively classified in a large database [10].
Hand geometry measurements are easily
collectible from both the hands. Palm is the inner surface of a hand between the wrist and the fingers. Palm print has been used as a powerful means in law enforcement for identification because of its stability and uniqueness.

In the first part of this review article,

we present an overview on some of the major research areas in palm print analysis. To set the scene, we first give a brief description of the biometrics background required for a proper understanding of the material.

IJSER © 2013 http://www.ijser.org

International Journal of Scientific & Engineering Research, Volume 4, Issue 7, July-2013 2069

ISSN 2229-5518

In the second part of this review article, we present an overview of the biometric technology and palm print technique analysis and comparison.

2. PALM PRINT BIOMETRIC SYSTEM

In the palm print biometric system there is three major steps are
• Acquisition of Palm Prints of all users in a image as shown in figure 1.
• Feature extraction for each class of palm prints and update of the database.
• Feature extraction of scanned input image.
• Matching with the stored features for the highest matching score to obtain the identification / verification output of the system.

FIGURE 1: SNAPSHOT OF THE PALM PRINT IMAGE

The verification system can be depicted as a block diagram as shown in figure 2. Feature extraction plays an important role in image identification and verification. There are many features exhibited in a palm. Hand images of every user can be used to extract the palm print. Alternately accurate palm print image is captured by a palm print scanner and then the AC signal is converted into a digital signal, which is transmitted to a computer for further processing. Some pre-processing may be necessary to bring the palm print images to a common coordinate system based on some hand geometry. A distance measure is used to measure the similarity of two palm prints in the palm matching process. Verification is regarded as a one-one matching. Identification on the other hand is concerned with the search for the best match between the training
sample and the test in the database, which is also termed as one-many matching.

Figure2. The proposed system of biometrics verification

3. FEATURE EXTRACTION

3.1. TEXTURE FEATURE AND INTERESTING POINTS -

Jane You and Wenxin Li have done careful study of new method to authenticate individuals based
on palmprint identification and verification. They propose a dynamic selection scheme to
facilitate the coarse-to-fine palmprint pattern matching by combining global and local
palmprint features in a hierarchical fashion. Here they detect the palmprint features as well as the
interesting points. These points are used as a process of registration of the palm print images
as in Figure 3. Datum point and Principal lines determination are done using directional
projection algorithm.

Fig.3. Definitions of a Palm print: Principal lines (1-heart line, 2-head line and 3-life line), Regions (I-finger-root region, II- inside region

IJSER © 2013 http://www.ijser.org

International Journal of Scientific & Engineering Research, Volume 4, Issue 7, July-2013 2070

ISSN 2229-5518

and III-outside region) and Datum points (a, b- endpoint, o-their midpoint).

Fig. 4. Track of datum point determination by using the directional projection algorithm Certain line geometry features such as slope and intercept are also used to match line segments between two palm images.
A verification function defined as-

Slope (i)=(y2(i) – y1)) / x2(i)-x1(i)). Intercept(i)=y1(i)-x1(i)slope(i),a(i)=tan-1(slope(i)).

Has been used for the final decision, (x1(i), y1(i)), x2(i), y2(i)), i=1, 2,...., N, where N is the number line segment. Each line segment can be represented by two parameters like slope and intercept.

In their experiment they have inked the palm- print on the papers and then scanned them to obtain 232×232 images. The 232 x 232 gray-scale inked palm-print images were taken with resolution of 125 dpi and digitised to 256 gray scales. Out of 200 distinct palm print images, 91 have been found to be in excellent agreement with the manual estimate as shown in table 1.

Comparison of different palmprint matching methods

Table. 1. Accuracy rate of 200 selected images
3.2. PERSONAL AUTHENTICATION BY PALMPRINT FEATURE – Chin-Chuan Han, Hsu-Liang Cheng, Chih-Chih Lin have used the scanner-based personal authentication by using the palm print features. This method is very suitable in many network based applications. They extract the meaningful features from the segmented ROI. They employ the simple Sobel operators to extract the feature points of palm prints. Four directional Sobel operators like S0, S90, S45 and S135. This operation is operated following expression:

f * S = max(f * S0, f * S45, f * S90, f * S135)

Fig.5. the generation of region of interest (ROI). Here the symbol * is defined as the convolution
operation. They applied Sobel operator and
present other complex morphological operators to extract the palm print features.
In their experiment a palm print input device with three kinds of window sizes 32x32, 16x16, 8x8
are adopted to evaluate the performance of the template matching methodology. They have been
used to acquire the palm prints of images with
845 x 829 pixels. An image has been collected

from 1000 positive samples of individuals x 49 times. A high recognition rate of 98% has been achieved both FAR and FRR values of matching scheme as shown in table 2.

IJSER © 2013 http://www.ijser.org

International Journal of Scientific & Engineering Research, Volume 4, Issue 7, July-2013 2071

ISSN 2229-5518

Table.2. Eigen Palm test results of 1000 images

3.3. LINE SEGMENT HAUSDORFF DISTANCE – given by Jingying Chen and Maylor K. Leung (2002) considers palm print as a new approach has been advantage to incorporate structural and spatial information to compute dissimilarity between two sets of line segments. They proposed the line segments Hausdroff distance generated from logos with three modules like segmentation, logo representation, and matching results. In this method the two sets of points M = {m1, .... , mp}(representing a model in the database) and N = {n1, .... , nq}(representing a test image), the Hausdorff distance is defined as

H(M,N) = max(h(M,N),h(N,M)),

The function h(M,N) is called the directed Hausdorff distance from M to N. It identifies the point of M that is the farthest from its nearest neighbours in N. Thus the Hausdorff distance, H(M,N), measures the degree of mismatch between two sets. Intuitively, if the Hausdorff distance is d, then every point of M must be within a distance d of some point of N and vice versa.

Fig.6 A sample of line segment representation image

An illustration of segmentation: (a) intensity image, (b) edge image, (c) thinned image.

In their experiment Hausdorff distance is a shape comparison metric based on binary image. It is a distance defined between two sets of points. Here they have taken 105 distinct logo images in the TIFF format. 189 test logos, mostly generated from 20 of the 105 models, were used to test the system performance. The 20 logos were selected
evenly according to the number of lines detected on each model logo. The first test set 20 generated logos has been correctly recognized by the proposed system. Most test logos 99% have been classified correctly. They uses rotated and scaled logos from a database of approximately 100 logos.

Table. 3. The result with different threshold values

3. ANALYSIS AND CONCLUSION

A different approach is presented to authenticate individuals by using their image features. This article presents a review of applications of image processing to the emerging field of biometrics.

In this paper we have presented some of the
early work on palm print as a biometric identifier which has set milestones in this area.
Some of the issues in using these methods are – (1) Palmprint is one of the most unique,
reliable and stable personal characteristics
which provide a powerful means to
authenticate individuals for many security system.
(2) The principal lines of some persons may be identical. Some of the persons may have strong wrinkles and some of them have little or no wrinkles.

IJSER © 2013 http://www.ijser.org

International Journal of Scientific & Engineering Research, Volume 4, Issue 7, July-2013 2072

ISSN 2229-5518

(3) The lighting condition is a major issue for geometrical features and texture features.
(4) The orientation of the hand while acquiring the palm print could pose a problem in

feature matching
Table 6. Summarizes these methods.

4. REFERENCES

1. R. Clarke. Human identification in information systems, Management challenges and public policy issues, Inf. Technol. People 7 (4) (1994) 6-37.
2. A.K. Jain, R.Bolle, S.Pankanti, Biometrics: Personal identification in Networked Society, Kluwer Academic Publishers, Dordrecht, 1999.
3. B.Miller, Vital signs of identity, IEEE Spect. 32 (2) (1994) 22-30.
4. D.doermann, E.Rivlin, 1. Weiss,
Applying algeberic and differential
invariants for logo recognition, Mach. Vision Appl. 9(1996)73-86.
5. A. Soffer., H.Samet, Using negative shape features for logo similarity matching, proceedings of the Fourth International Conference on Pattern Recognition, Brisbane, Australia,
1998, pp.571-573.
6. Y.S.Kim, W.Y.Kim, contend based trademark retrieval system using a visually salient feature, image vision compute 16(1998)931-939.
7. G.Gortelazzo, G.A.Mian, G.vezzi, P.Zamperoni, Trademark shapes description by string matching techniques, pattern recognition
27(8)(1994)1005-1018.
8. A.K.Jain, A.Vailaya, shape based retrieval, a case study with trademark image database, pattern recognition
31(9)(1998)1369-1390.
9. B.M. Mehtre, M.S. kankanhalli, W. F.
Lee, shape measures for contend based imageretrieval, a comparision, inf.process manage, 33(3)(1997)319-
337.
10. G.Ciocca,R.Schettini, contend based similarity retrival of trademark using relevance feedback, pattern recognition 34(2001)1639-1655.
11. D.D. Zhang, Automated biometrics technologies and systems, Kluwer Academic publishers, Dordrecht,
2000.

IJSER © 2013 http://www.ijser.org

International Journal of Scientific & Engineering Research, Volume 4, Issue 7, July-2013 2073

ISSN 2229-5518

Kasturika B. Ray has a Post graduate degree

in MIT (Master of Information Technology), Manipal Deemed University, Karnataka and continuing her Ph.D. research in the field of Digital Image processing in Computer Science and Engineering, SOA University, Bhubaneswar, India under the guidance of Dr.

Mrs. Rachita Misra. She has nearly eight years of teaching experience.

Currently she is working as a Dean Academics in Eklavya College of Engineering Technology and Science, Kusumati, Bhubaneswar, Odisha, India. She has several publications in the areas of Image Processing in International and National journals, Seminars and Conferences. She has published three International Journal research papers and presented in two National conferences and has attended ten National Workshops / Seminars etc. She has one article. She is the research member of Global Research Club. Her area of interest is Digital Image Processing and Networking.

CV Raman College of Engineering

IJSER © 2013 http://www.ijser.org