A Self-Synchronized Image Hash Algorithm

Abstract

Image searching and authentication have become important issues for the digital world. Image hashing technique has been proposed to meet them. In this paper, a self-synchronized image hashing algorithm resistant to rotation is presented. In the algorithm, the features of the image are extracted by Scale Invariant Feature Transform(SIFT) algorithm. Then a cyclic matrix are generated from the features. Thus, the rotation is translated into the elementary transformation of the matrix. At last, the hash value is calculated by the eigenvalue decomposition. Because the eigenvalues are invariant to the elementary transformation, the rotation is compensated. Thus this algorithm is very robust to rotation attack. Benefited from the efficient image representation capability of SIFT, the robustness and discrimination of proposed method are improved. The experimental results prove that the proposed method is robust to the simple attacks.