Image authentication schemes verify the integrity of an image against malicious manipulations. Most image authentication methods treat all manipulations as attacks. However, with the increasing demand for protecting JPEG images, image authentication methods that distinguish JPEG compression from malicious attacks are increasingly needed. This work presents a DCT-based image authentication approach with attack recovery, which recovers the attacked blocks. Two quantization properties, DQP and FQP, are presented to embed authentication features. The embedded feature is tolerant to JPEG compression with a quantization step of less than twice the pre-determined quantization step, but is sensitive to other malicious attacks. The attacked blocks are indicated after verifying a protected image, and the embedded recovery features are then extracted to recover these attacked blocks. Moreover, in case of the damage of recovery feature, an edge-based interpolation recovery approach (EIRA) is proposed to improve recovery results. Experimental results show that the proposed approach efficiently detects and recovers attacked blocks.
International Journal of Innovative Computing, Information and Control 6(3A), pp.885-896