A Block Based Parallel ZM-SPECK Algorithm

Abstract

The zero memory set partitioned embedded block (ZM-SPECK) technique is an embedded and memory efficient image compression algorithm. However, it is computationally complex due to the repetitive significance checking of sets and coefficients in each and every bit plane. To overcome this limitation, it is proposed to parallelize the algorithm over smaller blocks to reduce the overall encoding and decoding times of ZM-SPECK algorithm. The proposed approach called block based parallel ZM-SPECK (BPZM-SPECK) decomposes the wavelet transformed image into independent nonoverlapping spatial blocks utilizing the unique child-parent relationships in spatial orientation trees (in wavelet domain) and concurrently encodes every single bits in each bit plane of a block. The experimental results show significant improvement in computation time over the existing ZM-SPECK algorithm.