Investigation of better portable graphics compression for iris biometric recognition

Abstract

In this paper we present the very first study on the effectiveness of the recently proposed Better Portable Graphics (BPG) image compression algorithm in the context of iris recognition. Original and pre-processed iris images of the IITDv1 iris database are compressed at various reasonable bitrates and the impact of BPG on recognition accuracy is estimated in a bilateral and unilateral compression scenario. In experiments we found that, compared to well-established image compression standards recommended for biometric data interchange, JPEG and JPEG 2000, BPG generally reveals the least impact on the recognition accuracy of two conventional feature extraction techniques. In addition, we observe that iris segmentation is least affected when employing BPG compression. Consequentially, we identify BPG as an adequate choice for image compression in iris recognition.