Abstract: How to ensure watermark security, how to overcome the contradiction between robustness and JPEG lossy compression, how to cancel out the contradiction between fragility and malicious tamper, and how to reduce the computational complexity are the main issues to be solved in the existing semi-fragile image authentication algorithms based on watermark technique. In this paper, a semi-fragile image authentication algorithm using Zernike moments and watermark is proposed. It can distinguish malicious modification from incidental modification according to semi-fragile characteristics of Zernike moments of the low frequency subband in the integer wavelet transform domain of an image. Combining the semi-fragile characteristics of Zernike moments and the watermark based on human visual system, it can discern counterfeit attack, thus, improving watermark security. Computational complexity is reduced by integer wavelet transform using lifting scheme. Experimental results show that the proposed algorithm tolerates JPEG lossy compression to a large extent, and meanwhile it is fragile to malicious tamper and capable of locating the tampered area accurately.