A Palmprint Classification Scheme using Heart Line Feature Extraction

Abstract

A new technique to classify palm prints is proposed in this paper. A rectangular region of interest (ROI) containing only the heart line is extracted from palm print images obtained from a peg-free scanner. The ROI extraction is robust using boundary tracing and rotations based on our study of palm geometry. Preprocessing operations such as intensity normalization and smoothing are applied. Sobel gradient thresholds are applied to extract the heart line from the ROI. The palm print images are classified based on the regions that the heart line traverses in the palm horizontally. Our scheme allows for a total number of 257 possible categories. Testing of the scheme on two databases shows that a classification accuracy of more than 98% is obtained. It is expected that this very efficient method shall be useful in the classification and matching of very large sized palm print databases.