International Journal of Computer Applications (0975 – 8887) Volume 55– No.16, October 2012 15 Palm Print Recognition using Zernike Moments Subhajit Karar Information Technology Jadavpur University Kolkata, West Bengal Ranjan Parekh School of Education Technology Jadavpur University Kolkata, West Bengal ABSTRACT This paper proposes an automated system for recognizing palmprints for biometric identification of individuals. Complex Zernike moments are constructed using a set of complex polynomials which form a complete orthogonal basis set defined on the unit disc. Palmprint images are projected onto the basis set resulting in a set of complex signals. The magnitude of the complex value is computed and a scalar value is derived from it by computing the mean of the vector elements. Classification is done by subtracting the test samples from the mean of the training set. The data set consists of 80 images divided into 4 classes. Accuracy obtained is comparable to the best results reported in literature General Terms Pattern Recognition, Computer Vision Keywords Zernike moment, Palmprint recognition, Texture classification 1. INTRODUCTION Biometrics refers to automatic recognition of individuals based on their physiological and behavioral characteristics, which are nowadays used extensively for personal identification worldwide. Different physical traits like face, iris, fingerprint, palmprint, retina etc. fall under the perview of biometrics. Palmprint recognition involves identifying an individual by the principal lines, wrinkles, ridges on the surface of the palm. The basis for using palmprints lies in the fact that no two individuals have exactly the same palmprint pattern, moreover palmprints remain more or less stable throughout the lifetime and are easily obtainable using standard imaging techniques. Challenges in palmprint recognition are related to building a reliable data model from randomly oriented irregular lines that enable high amount of accuracies in security based systems and applications. This paper presents an efficient algorithm for palm print recognition by utilizing complex Zernike moments. The organization of the paper is as follows: section 2 provides an overview of the related works, section 3 outlines the proposed approach, section 4 details the experimentations done and results obtained, section 5 analyses the current work vis-à-vis contemporary works, section 6 brings up the overall conclusions and future scopes. 2. RELATED WORKS Moments and functions of moments have been used as pattern features in a number of applications. Some of the earliest works include [1, 2]. Zernike moments are a class of complex moments known for their invariance properties. In [3] the authors propose a palm print verification system using high order Zernike moments with k-nearest neighbour (k-NN) classifiers. Total 625 images are used from Hong Kong PolyU palm print database. In this paper the reported accuracy is 98%. In [4] palm print recognition algorithm is based on Robust Oriented Hausdorff Similarity (ROHS) measure. Total 7752 palm images of Poly Technique university of Hong Kong is taken as dataset. In [5] the authors propose a contactless palm print principal line-based feature extraction technique. Here DFT technique is used to calculate distances from endpoints to endpoints and point of interception to endpoints. Correlation technique, power spectral matching and Euclidean distance are used as classifier. 100% accuracy is reported in this paper. In [6] a hierarchical multi-feature scheme is used. Two levels of features are defined: geometry feature based on distance (level-1 feature) and texture feature based on Zernike moment (level-2 feature). Classification is done with two different neural networks and the output is combined into one recognition system. In [7] both geometrical and palm-print hand features are used. Texture extractions are done using wavelet transform, 2D Gabor filter and derivative methods. Support Vector Machine (SVM) classifiers are used as identifier and verifier. In [8] an efficient algorithm is used to extract the Region of Interest (ROI). The images are made geometric invariant. In [9] feature extractions are done using Gabor transform and Genetic Algorithm based on the specific- user. The palmprint images are used from Biological Research Center of Hong Kong Polytechnic University (PolyU). In [10] a palm print identification is done using wavelet transforms. The wavelets used for the analysis are Biorthogonal, Symlet and Discrete Meyer. Total 500 images are taken as dataset. In [11] the authors propose an innovative touch-less palm print recognition system. Here local binary pattern (LBP) method is used for feature extraction. Classification is done using a modified probabilistic neural network (PNN). In [12] the authors propose a principal line based palm print verification technique. Here a modified finite Radon transform is used for feature extraction. Pixel-to-area comparison method is used as classifier. In [13] the authors propose a palmprint recognition method using multiple correlation filters. Two ways of edge detection of images are done here. Firstly computing edginess of image of the palm print and secondly using phase symmetry processing of the image the edge is detected. 385 classes are used as the dataset. In [14] three orthogonal moments namely Zernike moments, pseudo Zernike moments and Legendre moments are used for feature extraction from palmprint images. Pseudo Zernike moments of order of 15 produce best result. Using this moment’s accuracy is achieved 95.75%. In [15] higher order Zernike moments is used for hand geometry feature extraction and the Log Gabor filter is used for palm print feature extraction. Overall accuracy achieved by combining these two features. 3. PROPOSED APPROACH This paper proposes an efficient algorithm to recognize palm print images using Zernike moment with a scalar feature representation.