Semi-fragile watermarking for tamper detection

We present a novel semi-fragile watermarking scheme for the tamper detection and authentication in digital images. The proposed algorithm firstly generates a transparent, robust watermark that sensitively depends on a visually meaningful binary watermark and on the content features in the frequency domain of an image. Then the image is divided into blocks and the transparent, robust watermark is embedded into the discrete wavelet transformation (DWT) domain of the image per block by quantizing the corresponding coefficients with public keys. At the same time, we propose a rule to distinguish the patterns of tamper between mild distortion and severe distortion. The experimental results demonstrate the watermarked image generated with the proposed algorithm can apply to JPEG lossy compression to a certain degree, and accurately detect and locate where the image has been suffered from malicious tamper for different resolution.