A novel direction-adaptive wavelet based image compression

Directional wavelet can effectively capture the directional dependence in images. However, the computational complexity is high. Based on the image statistics estimated by the structure tensor, a novel directional lifting image coder locally adapting the filtering directions to image content is presented. Before performing wavelet transform (WT), the proposed algorithm detects all the image blocks in a given image to decide whether the block is homogenous or not. For homogeneous block, the conventional 2-D discrete WT is used. This will considerably reduce the computational complexity and the number of bits needed to code the directional information. On the other hand, heterogeneous block is decomposed using directional lifting wavelet transform, which can effectively capture the directional dependence in the selected image and improve the coding gain of the image coder. Experimental results have shown that, compared to some existing methods, the proposed scheme has a better performance in terms of peak signal to noise ratio (PSNR) and subjective quality.