Research And Application Of Image Authentication Technology Based On Perceptual Hashing

In the field of image authentication,with the increasing complexity of digital image processing technology,image content,image processing operations,and image tampering types are developing in a variety of ways,which leads to many urgent problems in image content authenticity,integrity,and security authentication.Perceptual hashing is a key technology for multimedia security applications.Image perceptual hashing uses the method of image feature extraction to generate hash codes,and realizes fast image authentication by comparing the distance between hash codes of different images.This technique can not only effectively defend against most image processing attacks,but also improve the efficiency and accuracy of authentication.The perceptual hash algorithm can use the SIFT algorithm proposed by David Lowe et al.in the feature extraction stage.Because the SIFT algorithm uses a fixed contrast threshold for key point screening,the ability to detect key points in low-contrast images is limited,and the number of detected key points is too small.At the same time,the key point descriptor generated by the SIFT algorithm is too large in dimension,which leads to a large time complexity in the subsequent key point matching process.Aiming at these problems of SIFT algorithm,this paper proposes an improved SIFT algorithm,which uses adaptive contrast threshold to filter key points in the key point detection stage;uses LLE algorithm to reduce the dimensionality in the key point descriptor generation stage.The test results show that the improved SIFT algorithm has greatly improved the detection ability of key points in low-contrast images,the matching accuracy between key points,and the matching speed compared with the original SIFT algorithm.The SIFT algorithm is based on single image robust feature extraction,which is insufficient in describing image content.To solve this problem,this paper proposes an improved perceptual hash image authentication algorithm,which combines the global feature extraction based on Zernike moments and the local feature extraction based on SIFT key points to enhance the description of image content and make it suitable for different image processing operations.have wider robustness.The test results show that the algorithm proposed in this paper has good robustness to conventional image processing attacks,and has good discrimination against different images and tampered images.The comprehensive ROC performance is better than SIFT-PCA,SIFT-DWT and based on Hash algorithm for CNN feature clustering.This thesis applies the proposed Zernike-SIFT perceptual hash image authentication algorithm to the blockchain,and designs a digital image storage and authentication application scheme based on the Ethereum blockchain.Through the smart contract,the perceptual hash code of the image and the transaction details are stored in the Ethereum blockchain to realize the effective verification of the authenticity and integrity of the digital image,providing a solution for image authentication.