A zero-watermarking image authentication scheme using Zernike moment and extreme learning machine

Based on Zernike moment and extreme learning machine (ELM), a lossless watermarking algorithm against various attacks called zero-watermarking is addressed in this paper. Firstly, image normalization and lifting wavelet transform are used for the invariance of translation and scaling and suppressing of noise. Then Zernike moment magnitudes of training sample images and watermarking image are utilized to construct ELM training modle, whose memory ability enhances the scheme´s performance of resisting attacks. Experimental results prove that the proposed algorithm possesses strong robustness to common signal processings and geometrical distortions.