Image characteristic adaptive image data approximations based on human visual systems

Adaptive approximation methods are proposed based on local image characteristics and human visual systems. Image data is approximated by evaluating human sensitivity to the differences between pixels and the background, as well as the spatial frequency of the local area. Also, an approximation algorithm is proposed based on tailing zeros of image data with brightness sensitivity. Image data approximation is maximized by combining the proposed multi-algorithms. Experimental results show 43.77% average image data reduction while maintaining SSIM 0.98 with the Kodak dataset.