Image-aware Hashing Algorithm And Its Application Research

With the rapid development of multimedia,Internet,big data and other technologies.In particular,5G technology and full coverage of smartphones.It has ushered in a new era of science and technology and social development.Multimedia information exists in every aspect of human life.People can access and disseminate multimedia information anytime and anywhere.The authenticity and copyright protection of multimedia information have aroused people's wide concern,among which the image information is most used.Due to the use of lossy compressed image format on the Internet,the image authentication technology based on traditional hashing technology is no longer applicable.In view of this,researchers proposed a new image processing technology based on digital watermarking technology,multimedia authentication technology and traditional hashing technology--image perception hashing technology.Perceptual hashing technology establishes the mapping between the perceptual information level and its perceptual abstract.It makes objects that look similar have similar perceptual hashing values.It is robust to conventional signal processing operations and sensitive to tampering operations of various perceptual contents.As a new research direction,perceptual hashing technology can be used in image content authentication and authentication,content indexing and analysis,digital watermarking,image retrieval,and copy detection.This paper mainly does the following work:First of all,the origin,research background and significance of image perception hash algorithm are described.The domestic and foreign research achievements in this field are deeply studied and classified.The characteristics of each algorithm are simply described,and the advantages and disadvantages of different algorithms are analyzed.Secondly,the basic knowledge of image perception hash is introduced,including definition and properties,common data sets,similarity measurement methods and algorithm performance evaluation indexes.Then,two new perceptive hashing algorithms are proposed to solve the problems that are not well solved by the current algorithm.One is the perception hash algorithm based on semi-supervised learning and adaptive threshold,which is applied to image tampering detection.This algorithm USES semi-supervised learning method to The semi-supervised learning method is used to encode image features cooperatively in a compact public binary code space and sets adaptive threshold,which further improves the classification performance of hash algorithm.The other is a robust image hash method based on ring segmentation and color layout descriptor,which can resist rotation operation by extracting color layout descriptor features from the ring and generating a robust hash by compression.At last,the paper summarizes the content of the whole paper,and points out the further research direction in view of the existing problems of the current hashing algorithm.