Adaptive coefficient sorting and selection for high capacity data hiding in JPEG images
摘要
Considering the widespread utilization of the Joint Photographic Experts Group (JPEG) in various applications, some studies have recently employed JPEG images as a medium for data hiding. This is accomplished to conceal data within images, consequently reducing the risk of data leakage. While earlier studies have demonstrated remarkable performance, they have not sufficiently addressed the embedding efficiency of Alternating Current (AC) coefficients. Overlooking this aspect results in greater image distortion after embedding. To optimize embedding, coefficients with smaller errors are given priority over those with larger errors. A threshold is then established to restrict the embedding of coefficients that exhibit higher errors. Finally, non-zero AC coefficients within each coefficient block are concatenated and paired to encode data. Compared to previous methods, with a quality factor of 90 and a payload of 5000 bits, the average image quality improved by 1.89 dB across 200 test images. Additionally, the average file size decreased by 77% for the six test images.