Region-based tampering detection and recovery using homogeneity analysis in quality-sensitive imaging

This paper presents a region-based tampering detection and restoring scheme that exploits both lossless (reversible) data hiding and image homogeneity analysis for image authentication and integrity verification. The proposed scheme enables the detection of a tampered region and then recovers it by using the embedded information selected as the recovery feature. Moreover, since reversibility helps in taking the correct decision during image analysis, it is highly desired in quality-sensitive imaging such as medical imaging and satellite imaging, where even minimal distortion introduced by embedding data is unacceptable. To extract the recovery feature, we analyze the image homogeneity using quad-tree decomposition, which adaptively divides a square image into several variable-sized blocks, and choose the average value of each block as the feature. This makes the length of the feature more efficient while ensuring that the visual quality of the restored image is better than that in the case of the conventional fixed-size block-based approach. Through experiments on 8-bit, 12-bit, and 16-bit images, we demonstrate the effectiveness of tampering localization of the proposed method and show that restoration is achieved with high visual quality.