Min Max Threshold Range (MMTR) based approach in palmprint authentication by sobel code method

Abstract

Palmprint recognition is an effective biometric authentication method to automatically identify a person’s identity. Palmprint is rich in features like geometry features, line features, datum points, delta features and minutiae features. Several edge detection methods are available to extract line feature from the palmprint. In this paper, the hand image is pre-processed to get the desired Region of Interest (ROI)/palmprint. Multiscale Sobel Code operators of different orientations (0°, 45°, 90°, and 135°) are applied to the palmprint to extract Sobel-Palmprint features in different directions. The Sobel-Palmprint features extracted are stored in Sobel-Palmprint feature vector and matched using Hamming Distance similarity measurement method. In addition, a Min Max Threshold Range (MMTR) method is proposed that helps in increasing overall system accuracy by matching a person with multiple threshold values. In this technique, firstly the person is authenticated at global level using Reference threshold. Secondly, the person is authenticated at local level using range of Minimum and Maximum thresholds defined for a person. Generally, personal authentication is done using reference threshold but there are chances of false acceptance. So, by using the Minimum and Maximum Thresholds range of false accepted persons at personal level, a person is identified to be false accepted or genuinely accepted. MMTR is an effective technique to increase the accuracy of the palmprint authentication system by reducing the False Acceptance Rate (FAR). Experimental results indicate that the proposed method improves the False Acceptance Rate drastically.