Effects of image lossy compression on palmprint verification performance

Abstract

In this paper, we investigate the effects that the lossy image compression has on the performance competitive codes of palmprint verification algorithms. Effects of lossy compression on palmprint recognition performance are of interest in applications where image storage space and transmission time are of critical importance. To reconcile that goal with its implications for bandwidth and storage, we test the effects of image lossy compression on competitive code under different transforms, quantization and encoding algorithms using a publicly available palmprint database. Experiments performing recognition directly in the compressed domain show that the extracted feature template variation from the compressed palmprint image is not consistent with the image distortion. What is interesting, the effect on the JPEG is more serious than that of JPEG2000 when the compression ratio is bigger than about 11 while the effect on JPEG2000 is more serious than that of JPEG when the compression ratio is smaller than about 11. And the trellis coded quantized method is not suitable for compressing palmprint images.