Image Hashing based on Shape Context and Speeded Up Robust Features (SURF)

Image Authentication is one of the key issue in today's age of multimedia technology. With the availability of many hacking techniques and picture editing tools like Photoshop, image security is a major research field. Local feature points are elaborately used in resolving many problems, i.e., object detection, robust matching etc., but its use in the field of image hashing is not explored. In this work, we propose image hashing based on shape-context via Speeded Up Robust Features (SURF). The work is motivated by SIFT-based approach. Our contributions are: 1) SURF based algorithm is several times faster than SIFT-based algorithm. 2) It is more robust against different type of image modifications than SIFT. 3) We have incorporated rotation invariance property. Rotation is a difficult content preserving operation to model. The experiments are carried out on 1000 images and have shown that the proposed image hashing technique is robust against content-preserving operations. The region of convergence shows the efficacy of the proposed method compared to many state-of-the-art methods.