A Robust Blind Image Watermarking Scheme Based on Classified Vector Quantization

Since 2000, many digital watermarking schemes for vector quantization (VQ)-compressed images have been proposed. Their main idea is to carry watermark information by VQ codeword indices. The advantage of this kind of watermarking schemes is its robustness to VQ compression with the same codebook. This Letter presents a more effective image watermarking method based on classified VQ. First, the input image is segmented into blocks and a classifier is used to classify each block. Then, each block is encoded by the classified VQ codebook according to its class and the two watermark bits corresponding codebook. Experimental results show that the proposed scheme is a blind watermarking method which outperforms existing VQ-based watermarking methods, and the watermarked image is robust to most of common signal processing operations.