Personal Classification Using Palmprint by Exploiting Blackman Filter

Palm print based personal verification is considered to be the most accurate and precise in comparison to all other modalities in biometric domain. It has a high user satisfactoriness and reliability. This makes the palm print modality to be very useful in the application of information security, ATMs, and physical access etc. The palm print recognition technology is easy to use and hard to counterfeit because the attributes are not something that can be misplaced or lost, easy or conveyed from one person to another. It can also be combined with other biometric modality like finger print to form a robust and reliable biometric verification system. With palmprint recognition technology becoming more commonplace, naturally there is much discussion as to the effectiveness and accuracy behind it. Palmprint texture analysis incorporates not only the analysis of principal lines, wrinkles, minutiae, delta points, ridges but also assimilates roughness or smoothness of palm. This paper presents a neoteric approach for palmprint based personal identification which exploits the textural pattern of palm print using the Blackman Filter. In order to extract the Region of Interest (ROI), second order statistical moments have been utilized to obtain the parameters of best fitting ellipse in which the major axis of the ellipse corresponds to the longest line in the image and was assumed to be passing through the middle finger. Ratios between eigen values help examine the shape of an object whereas direction of elongation is evaluated using the direction of the eigenvector corresponding to highest Eigen value. Henceforth, the offset between the normal axis and the major axis of the ellipse is calculated. After determining the ROI, the two-dimensional (2-D) spectrum is decomposed into fine slices using iterated directional filter banks. Subsequently, directional energy component for every block from the sliced sub band output is calculated. The proposed algorithm can capture both global and local details in palm print as a condensed fixed length palm code. The algorithm has been tested on 500 images of palm indigenously acquired. Normalized Euclidean distance is employed as an evaluation function for classification of palm images. Blackman Filter based approach established the decidability index of 2.7874, Equal Rrror Rate (EER) of 0.2454% and Genuine Identification Rate of 93.6%. The quantitative measures confirm palm print recognition to be reliable and accurate with Blackman Filter based approach as compared to one reported in literature.