Human visual system based data embedding method using quadtree partitioning

In this paper, we proposed a data embedding method based on human visual system (HVS) and quadtree partitioning. For most HVS-based methods, the amount of embedded data is based on the measurement of differences of pixel pairs or the standard deviation of image blocks. However, these methods often result in larger image distortion and are vulnerable to statistical attacks. The proposed method employs a specially designed function to measure the complexity of image blocks, and uses quadtree partitioning to partition images into blocks with different sizes. Larger blocks are associated with smooth regions in images whereas smaller blocks are associated with complex regions. Therefore, we embed less data into larger blocks to preserve the image quality and embed more data into smaller blocks to increase the payload. Data embedment is done by using the diamond encoding technique. Experimental results revealed that the proposed method provides better image quality and offers higher payload compared to other HVS-based methods.

► Images are partitioned into blocks using quadtree decomposition.
► The block complexity is measured along the Hilbert curve.
► Messages are embedded according to the measured complexity.