Perceptual Image Hashing Technology And Its Application In Digital Authentication

Digital images are becoming widely used in our daily life as a kind of common multimedia information. Therefore, the authentication of their authenticity and integrity has become increasingly prominent. Traditional hash functions e.g. MD5 or SHA-1, whose purpose is to extract a short binary string as a digest from a large digital message, are frequently called message digest functions and extensively applied to the authentication of the authenticity and integrity of text data. Their key feature is that they are extremely sensitive to the input message; i.e., changing even one bit of the input message will change the output dramatically. However, image data often undergoes various manipulations of preserving content such as lossy compression, filtering and noise pollution etc. So the image hash values should depend on image contents. Because of the above problems, this paper deeply researches the perceptual image hashing technology from the perspective of robust and security. At the same time, in the traditional digital identity authentication system, because of the uniqueness of biometric features, even if people have known that they leaked, they also can't modify them, which will lead to seriously secure problems. So this paper introduces the perceptual image hashing technology into the digital identity authentication system and provides a novel method to solve the problems.The main contents of this paper are showed as follows.1,Two classical perceptual image hashing algorithms are realized and an algorithm based on invariant moments is proposed. Normalized Hamming distance is used to measure their performance. Experimental results indicate that this algorithm can resist the modification of preserving content, insure the security under the control of the secret key, and has the best robustness in comparison with the two classical algorithms.2,An algorithm based on improved Hausdorff distance and feature point extraction is proposed, which overcomes the lacks of existing algorithms which cannot resist the attacks of image enhancement, printing and scanning, large angle rotation. Improved Hausdorff distance is used to measure its performance. Experimental results indicate that this algorithm not only has strong robustness but also can detect whether an image is tampered or not.3,The image hashing technology is introduced into the digital identity authentication system, a multi-factor digital identity authentication method based on image hashing technology is proposed and a multi-factor authentication system is constructed. In this way, security problems in traditional biometric authentication systems can be solved. Experiments with common face, fingerprint and iris databases show that the approach is quite effective.