Adaptive Quantization in JPEG2000 Framework
Rong Zhang · Jisuanji fangzhen · 2007
In this paper, an adaptive quantization strategy based on image texture classification is proposed. Based on the characteristics of the human vision, the new approach takes into account the statistical characteristics of each subband and the different effects of wavelet coefficients in each subband on saving the edge and texture information of original image. Each high frequency subband is divided into blocks. The blocks are classified as smooth blocks, edge blocks or texture blocks. The wavelet coefficient is quantized adaptively depending on which kind block and subband it belongs to. A specific application of the proposed strategy to JPEG2000 is presented. Experimental results show that this approach can save the edge and texture information well at low bit rate and the subjective quality of the reconstructed image is improved to some extent.