Résumé
In this paper, we consider the problem of the quantization stage of robust perceptual image hashing in authentication application. The main goal of our analysis is to study the behavior of the extracted image features to be hashed against malicious/non-malicious manipulations. In our analysis, we consider only an additive noise. Note that even if the noise is small, it can cause errors during the quantization stage of perceptual image hashing system. For this purpose, we present theoretical analysis of the extracted feature under additive noise, as well as the probability of a false quantization of these selected features. We then introduce our proposed approach based on this theoretical analysis. Finally, several experiments are conducted to validate our theoretical analysis.