Research On Multimedia Removal Based On Perceptual Hash

With the development of human information technology,The forms of information dissemination on the Internet are becoming more and more diverse.Throughout the Internet,the amount of data accumulated in various industries is getting larger and larger,and the digital content of enterprises and individuals is also growing infinitely.There is a lot of redundancy in multimedia data in the cloud,At the same time,through massive clients,upload new repeated data,wasting a lot of bandwidth.Therefore,how to get rid of again efficiently becomes an urgent problem to be solved.This paper focuses on the problem of multimedia data removal.Based on the theoretical analysis and algorithm research of perceptual hashing,a new algorithm for constructing perceptual hashing is proposed.At the same time,two multimedia deduplication schemes based on perceptual hashing are designed,which are image deduplication based on perceptual hashing and video deduplication based on perceptual hashing.The main findings are as follows:1.A novel perceptual hashing algorithm BDCT-hashbased on block DCT variation is proposed.The algorithm extracts DC coefficients and partial AC coefficients after discrete cosine changes in blocks of images,and constructs perceptual hash sequences by using the coefficients relationship between blocks.Through a large number of experiments,the algorithm is compared with three classical perceptual hashing algorithms,and it is proved that the algorithm has higher removal rate and lower error rate,and it is feasible.2.Aiming at the popular JPEG compression coding,the proposed BDCT-hash algorithm is improved.A fast hashing algorithm is proposed,which extracts correlation coefficients directly from JPEG image compression stream and integrates them into hash sequences,which reduces the time cost of the algorithm.3.Based on the proposed fast hash algorithm and the previous average hash algorithm,a new image deduplication scheme based on the hybrid hash algorithm is proposed.The scheme includes similarity detection,ownership authentication and image quality evaluation.Through a large number of experiments,the time cost of the accuracy of the scheme is analyzed,and the feasibility of the scheme is proved.4.In this paper,an adaptive threshold key-frame extraction method based on perceptual hashing is proposed,which is beneficial to the proposed perceptual hashing algorithm BDCT-hash.based on this,a video deduplication scheme is proposed,which includes key-frame extraction.Through the existing video quality evaluation algorithms,the video quality of the video candidate set is evaluated,and the video dereoperation is finally carried out.The experimental results show that the proposed scheme has a high accuracy.