Fingerprint compression using a modified wavelet transform and pyramid lattice vector quantization

A new compression algorithm for fingerprint images is introduced. A modified wavelet packet scheme which uses a fixed decomposition structure, matched to the statistics of fingerprint images, is presented. A technique for determining the most important coefficients is introduced. The algorithm uses both hard and soft thresholding schemes to make the procedure fast and efficient. The bit allocation for each subimage of the modified coefficients is determined. Each subimage uses a different quantization technique based on its entropy. Then, a lossless compression technique, Huffman, is used to obtain further compression. The algorithm results in a high compression ratio and a high reconstructed image quality with a low computational cost, compared to other existing algorithms. The performance of the proposed algorithm is compared to that of other decomposition techniques: ordinary wavelet transform (OWT), entropy-based best basis selection (E-BBB), wavelet/scalar quantization (WSQ) and JPEG.