| abstract |
The present invention provides a method, system and device for lossless compression and decompression of JPEG images, comprising: decoding a JPEG image part, obtaining DCT coefficients and quantization tables of luminance and chrominance, and buffering them; sorting the DCT coefficients according to corresponding frequency bands, Divide the sorted DCT coefficients into several frequency groups; for the frequency groups, use a convolutional neural network to extract contexts from the DCT coefficients of the frequency band to which they belong; according to the contexts, predict the probability distribution of the DCT coefficients, and perform arithmetic Encode to get a binary code stream. The present invention improves the compression rate of the JPEG image on the premise of only partially decoding the JPEG image, and does not change any content in the image. The present invention is compatible with the currently widely used JPEG image coding format. |