A non-uniform image compression using genetic algorithm

Most of the image compression standards are based on frequency domain transforms that is followed by quantization. Therefore, the ways to threshold a transformed block of image are the interest of many researchers in order to find a good balance between image quality and compression ratio. In this paper we proposed a lossy compression method based on JPEG which applies an adaptive threshold for each block. The threshold mask function is determined by using genetic algorithm. The proposed genetic algorithm uses gray level difference (GLD) as a feature. The experimental results showed that proposed method had higher PSNR than some related works and it is comparable with JPEG2000.