Improvement of Radon-Based Image Hashing Using Image Normalization

摘要

Image Hashing has several applications such as image authentication, image indexing and broadcast monitoring. In the content authentication applications, the hash values of perceptually similar images generated by content preserving modification, such as compression and moderate geometric distortions, also must be similar. Taking this into account we propose an improvement of the Radon transformbased image hashing algorithms, introducing an image normalization algorithm in the preprocessing stage. In the proposed scheme, robustness against geometrical distortions is increased considerably, while keeping the robustness to other content preserving modifications, such as compression, and filtering, which are offered by Radon transform.