Robust Image Hashing Using Histogram Reconstruction for Improving Content Preservation Resistance and Discrimination

This paper proposes a new image hashing method, which uses histogram reconstruction to solve the problem of not being sensitive change pixel position, while ensuring robustness algorithm against common content preservation attacks (such as blurring, noise addition and rotation). The proposed can resist arbitrary angles rotation, possibly because reconstructed leverages rotational symmetry its own invariance rotation operations. We measure similarity between different images by calculating Hamming distance hash vectors images. Our experiments show that method performs well in discrimination compared with other established algorithms. In addition, we conduct receiver operating characteristic curve analysis further verify superior overall performance our method.