Skip to Main Content
Binary image is first transformed, and then is embedded in the coefficients of subband of host image imperceptibly, which is divided into different parts, coefficients of band are found out based on adaptive algorithm. The extracting process is carried out by combining watermarking extracted from different sections of the image based on correlation coefficient between original watermark and derived one. The scheme proposed here by utilizing multi-address embedding and extracting based on adaptive algorithm is extremely robust to wide variety of regular cropping, and the derived watermark is still readily recognizable even if it is cropped randomly. Experiments show the good effectiveness of the algorithm normalized electronic documents in order for readers to search and read papers conveniently.