Research on adaptive restoration image fragile watermark algorithm
Lianhuan Li, Yao Yuxian, Yang Xinfeng · 2011
Combining the image spatial domain local properties with human visual system, the authentication data is acquired by computing class membership degree of each image block properties, which is generated by applying k-means clustering technique to cluster all the image blocks. Then the recovery data is composed of the average intensity by truncating the two least significant bits(LSBs) of each pixel with in each block. Finally, the Logistic chaotic encrypted feature watermark comprising 2-bit authentication data and 6-bit recovery data of image block is embedded into the two LSBs of each Pixel within its corresponding mapping block. The algorithm not only achieves superior tamper detection and locates tiny tampered positions in images accurately, but also has good capability of recovery the tampered regions effectively.