Data hiding techniques for printed binary images

The objective of this research is to develop a method to hide information inside a binary image by digital halftoning techniques with certain modifications. Two modified digital halftoning techniques, modified ordered dithering and modified multiscale error diffusion, are used in this research. The data is encoded pixel by pixel in the halftone image according to position at the image and sequence of binarization, respectively. The eye model and mean square error are used to measure the image quality. A computer vision method has been developed to recognize the printed binary image. The results show that thousands of binary images similar to human vision but quite distinct from each other by computer vision can be generated. The eye model and computer vision are useful for both binary image quality measurement and data recognition. These new techniques have great potential in printing security documents such as currency, ID card as well as confidential documents