Image tamper detection and recovery using adaptive embedding rules

• We propose a fragile watermarking scheme using differential embedding strategy.
• The proposed scheme has less damage to the original image.
• This scheme has good performance in authentication error rate and image quality.
• The proposed scheme is very applicable to low variation images.

Many image tamper detection methods implement authentication and recovery in units of blocks; however, these methods do not consider the characteristics of blocks to distinguish watermark embedding and detection modes, thus resulting in poor hiding effects. We suggest that embedding an excessive number of bits within a region with only a slight change in an image to record recovery information is unnecessary. In addition, more recovery information is required in a region with major changes to improve recovery quality. Therefore, this study used smoothness to distinguish the types of image blocks, and employ different watermark embedding, tamper detection, and recovery strategies for different block types to enhance hiding efficiency, authentication, and recovery effects. The experimental results regarding the authentication error rate and image quality showed that the proposed scheme has satisfactory performance.