Introduction - If you have any usage issues, please Google them yourself
Cosine transform image compression: DCT NN first image is divided into blocks of pixels as a whole (usually N = 8, i.e., 64 pixel blocks), and then each of the NN pixel block DCT transform. Since most high-frequency component image is small, the high frequency component corresponding to the image distortion less sensitive, more coarse quantization can be used, quality assurance in FIG required to discard some of the secondary information.