Embedded Multiple Subbands Scaling Hybrid Coding Based on Integer Wavelet Transforms

摘要

The Integer Wavelet Transform (IWT) allows both lossy and lossless compression using a single bitstream. However, its rate-distortion performance is inferior to the Discrete Wavelet Transform (DWT). In this paper, a new coding algorithm called multiple subbands scaling hybrid coding (MSHC) is presented based on the analysis of IWT signal characterization. The new algorithm scales each subband with the scaling factor, which can optimize the transform coefficient distribution of the subband. A segmentation threshold based on the integer powers of two is selected. All coefficients above this threshold are encoding using simple fast quadtree partitioning scheme. The residual coefficients below the threshold are encoding using block arithmetic coding. Experimental results show that the MSHC is able to raise the IWT lossy compression performance up to the DWT level and provides PSNR up to 0.4～0.8 dB better than JPEG2000 method using IWT.