Lossless Data Hiding with Quadtree Decomposition

In this paper, we present the loss less data hiding algorithm based on the concept of quad tree decomposition. With the slight changes of original images, by altering the shape of histogram, or by modifying the difference value between two consecutive pixels, loss less data hiding can be achieved at the encoder. Correspondingly, at the receiver, both the original image and the embedded data should be perfectly recovered. In this paper, we focus on utilizing inherent characteristics of original image based on quad tree decomposition, where smooth regions are partitioned with larger square blocks, and active regions are composed of smaller square blocks. With the different characteristics, different amounts can be embedded into blocks with different sizes in quad tree decomposition, in order to look for better quality of marked images both subjectively and objectively. Simulation results with proposed loss less data hiding algorithm based on quad tree decomposition have pointed out the better performances over relating schemes in literature.