Abstract

This paper presents a new iris feature descriptor, which first uses 2D-Gabor filter to extract energy orientation of iris texture before division, and then presents filtering distribution about the features within a block in each orientation statistics histographically. Thus, the orientation of feature-by-point energy is converted into that of histogram feature-by-block, during which the basic characteristics of the intra-class are maintained whereas the differences in inter-class are greatly widened. Such process largely improves distinguishability. Finally, the Euclidean distance is adopted for recognition. Experimental results on CASIA-V1 iris database show that this method has the advantage of a higher recognition performance over Daugman's, with the correct recognition rate up to 98.49% and the equal error rate down to 1.53%.

Paper Details