A Tampering Detection Algorithm Based on Multidirectional Authentication

As an important part of digital watermark technology, semi-fragile watermarking technology can not only effectively detect the tampered area, but also tolerate some conventional signal processing without changing the main content. In order to improve the tampering detection accuracy of semi-fragile watermarking algorithm, this paper proposes a multidirectional authentication strategy. The scrambled bit planes of image feature are embedded into multiple mapping blocks. When tampering occurs, some equations of the image feature and the embedded feature are established for the tampering localization. Based on the proposed multidirectional authentication strategy, a semi-fragile watermarking algorithm is proposed. The algorithm extracts the maximum singular values of image block as feature, and then encodes the feature by multidirectional scrambling to obtain the authentication watermark. In order to improve the robustness for JPEG compression, the authentication watermark is embedded into the wavelet intermediate frequency coefficients of the image by significant difference quantization. The experimental results show that the proposed algorithm has excellent ability of tampering localization.