Optimal reconstruction design for JPEG-coded image using structural similarity index

JPEG is used in image compression field widely. Due to the presence of quantization noise, image cannot be reconstructed perfectly. As we all know that discrete cosine transform (DCT) can be equivalent to subband coding system, and the IDCT matrix can be regarded as the polyphase matrix of synthesis filter. In this paper, we design the diagonal coefficient matrix K, which adjust the gain of synthesis filter so that the structural similarity index (SSIM) between the original image and the reconstructed image is maximized. And the SSIM is used as the optimization index to evaluate the performance of our scheme. The effectiveness of our method is shown in the experiment section.