Improved Fingerprint Compression Technique With Decimated Multi-Wavelet Coefficients for Low Bit Rates

Abstract

In this paper, a multi-wavelet transform with decimated frequency bands is proposed to be used in the Set Partitioning in Hierarchical Trees (SPIHT) algorithm to improve fingerprint image compression. Either shuffled or unshuffled multi-wavelets can be used for SPIHT algorithm. In both the cases, the quality of the compressed images at lower bit rates either remained the same or slightly improved compared to wavelets. To improve the performance at lower bit rates, a method which utilizes the decimated version of multi-wavelet for the initialization of lists in SPIHT algorithm is used. The multi-wavelet used for the proposed work is SA4 (Symmetric-Antisymmetric). The algorithm was tested and verified using NIST, Shivang Patel, NITGEN and other databases. An overall improvement in performance particularly at lower bit rates (0.01 to 0.09) compared to a multi-wavelet without decimation was obtained using this method. The improvement was 0.798dB, 0.857dB and 0.859dB for the images in NITGEN database for a multi-wavelet decimated by 2, 4 and 8 respectively. Similar performances were observed for other databases. It was further observed that the PSNR was highest when the multi-wavelet was decimated by a factor of 4.