A Robust Binary Text Digital Watermarking Algorithm for Print-Scan Process

Considering the simple and weak structure of binary texts, this paper proposes a novel algorithm which combines the methods of Block Partition, Discrete Cosine Transform (DCT) and pixel flipping. Based on appropriate block partition, the matrix of each block\´s characteristic values is transformed into the DCT domain, and a DCT coefficient matrix generated. Then, some high-frequency DCT coefficients are modified according to a threshold. Depending on the change of characteristic values before and after the Inverse Discrete Cosine Transform (IDCT), we can manipulate "flippable" pixels to embed an amount of data without causing noticeable artifacts. The hidden data can be extracted without using the original image, and can also be accurately extracted after high-quality printing and scanning, even reprography.