科研之友
微信
新浪微博
Facebook
分享链接
摘要
A novel semantic compression based on vector quantization (VQ) and data hiding is proposed. A compact version of the original image is generated, and then, VQ encoding is used in both the original image and the compact image to get their indexes of codewords in each block. In this processing, principal components analysis (PCA) is employed to rearrange the codebooks of the original image and the compact image in order to obtain the similar index table. Finally, the difference values of these indexes are calculated and embedded into the compact image using the reversible data hiding scheme to generate a small-sized compressed image that has similar content. And a high-quality reconstructed image with the original size easily can be obtained. Our experimental results have proven the expected merits of the proposed scheme.
