An improved Hu moment invariants based classfication method for watermarking algorithm

A novel semi-fragile algorithm incorporated with the Hu moment invariants in the wavelet-based contourlet transform (WBCT) is proposed in the paper. The invariance of parent and children coefficients relations is utilized for watermarks embedding is WBCT domain. Then the property of Hu moment invariants technique is adopted for classifying the malicious and non-malicious regions of manipulations when authenticating the suspected image. In our proposed semifragile watermarking scheme, the use of Hu moment invariant technique is not only has an excellent robustness to nonmalicious manipulation (JPEG/JPEG2000 compression, additive Gaussian noise and Gaussian filtering) but it also provides an effective fragility to malicious manipulations such as content copy& paste attack. Simulation results demonstrated that our proposed algorithm can localize the tampered region with higher detection rate.