Colour-based palmprint verification - An experiment

This paper presents biometric verification experiments based on palm colour information. Feasibility of colour components from several different colour models (RGB, normalized rgb, HSL, YUV, CIE XYZ, CIE LAB and Hunter LAB) for the purpose of biometric recognition was determined. According to the recognition rate, feature vectors based on normalized r and normalized b colour components are selected. Fusion at the matching score level is used to improve verification accuracy. Experimental results suggest colour can be used as a highly discriminatory characteristic, which in combination with other palm characteristics (principal lines, palm geometry, appearance-based features) enables the design of a biometric system for medium/high security applications.