An Improved Histogram-Based Image Hashing Scheme Using K-means Segmentation

The perceptual image hashing is an emerging technique which can be used in image authentication and content-based image retrieval. Design of an image hashing scheme which has both robustness and fragility properties is becoming a new challenge. Recently, Xiang et al. proposed a histogram-based image hashing scheme which achieves a good robustness to perceptually insignificant attacks. However, the fragility to malicious attack is not optimistic. In this paper, we propose an improved histogram-based image hashing scheme by using K-means algorithm, which obtains a better fragility result than original one. In particular, we firstly segment the image into two segments by using K-means algorithm, then extract histogram in each segment. Finally, we combine two bin groups and generate the image hash value. The experimental results confirm that our improvement not only preserves the robustness of original one, but also enhances the fragility to malicious attacks.