关键词: 图像哈希, 超复数Fourier变换, 对数极坐标变换, 位置扰乱机制, 真伪认证, 汉明距离

Abstract:  In order to solve these defects such as low detection accuracy and sensitivity of tampering contents, which are induced by converting three components R, G, B of the color image into grayscale image for achieving image content authentication without considering the chrominance information in current image Hash algorithm, this paper proposes the Hash authentication algorithm for color image based on super-complex Fourier transform coupled with polar coordinate mechanism. Firstly, the input image is preprocessed based on the mean filter and the rotation factor to enhance the algorithm robustness of image scaling, so as to minimize the effect of image information loss on algorithm. Then the log polar coordinate transformation is introduced to transform the pretreatment image into the log-polar field. And the super-complex discrete Fourier transform is constructed by super-complex theory and Fourier transform to extract image features with anti-rotation tampering. The image Hash model is constructed according to the amplitude coefficients of the super-complex discrete Fourier transform. Meanwhile, the position disturbance mechanism is designed based on Logistic chaotic map, which randomly scrambles the image Hash for enhancing the security of Hash model. Finally, by using a cryptographic Hash model and the Hamming distance, the similarity of the two Hash values between the original image and the user received image is calculated. The experimental results show that, for processing color image, this algorithm has stronger robustness and security, it can effectively resist the color, contrast, rotation, noise and scale tampering attacks compared with the current image hashing algorithm.

Key words: image Hash, hyper-complex Fourier transform, log-polar transform, position disturbing mechanism, true and false information authentication, Hamming distance