Partitioning morphological wavelet image coding

Recent successes in wavelet image coding are mainly attributed to the exploitation of within and cross subband redundancies with optimized rate-distortion analysis. There have been several very competitive wavelet coders developed, such as Said and Pearlman's SPIHT, Xiong et al.'s SFQ and the JPEG2000 standard etc. A novel wavelet image coding algorithm, namely partitioning morphological wavelet image coding (PMWC) is proposed in this paper, which extends MRWD by logically partitioning the information in the morphological dilated cluster and exploiting the within and cross subband dependencies in the significant coefficient areas. Extensive experiments have shown that the proposed PMWC outperforms SPIHT and MRWD, and is very competitive with the state-of-the-art codecs such as SFQ and JPEG2000. Because of its inherit joint compression and noise reduction properties, the reconstructed images often show somewhat superior perceptual visual quality.