Reversible Data Hiding Research Based On Single And Multiple Histograms Shifting

The information age is constantly advancing,so that people can quickly and conveniently obtain digital information from various channels.At the same time,multimedia information security is facing unprecedented challenges.Reversible data hiding(RDH)technology can hide secret data into the carrier without destroying the carrier and secret information during extraction.This feature makes RDH technology widely used in various fields of information security.Histogram shifting(HS)is a common RDH method,which is widely studied because of its high loading capacity and low distortion for origin image.Most of the existing RDH methods based on HS use the smooth region of the carrier image to construct a sharp histogram,so as to reduce the distortion caused by data embedding.This paper takes HS technology as the research object,which is divided into two technologies:single histogram modification(SHM)and multiple histograms modification(MHM).The existing RDH algorithm for single histogram is usually used to calculate the texture complexity(single feature)of image block,then divide the image texture region by the given threshold,and then construct histogram by using smooth area.However,most of the existing algorithms can not distinguish the texture region effectively,which leads to the distortion of the dense image.For MHM technology,the multiple features extraction scheme of one image block are meticulously designed.Meanwhile,according to similar features,the image blocks are divided into different types by utilizing the clustering algorithm,so as to construct the multiple histograms.However,the existing algorithms use the method of "violent iteration" to select embedding points for multi-histogram.The calculation complexity of this method is too high to greatly reduce the efficiency of the algorithm.The main contributions include:(1)For the RDH scheme with SHM,this paper proposes a RDH technology based on image edge detection algorithm,aiming at effectively dividing the texture area of image and reducing the distortion of the stego.In addition,a new method of computing complexity is proposed.The method can fully and accurately extract the relationship between pixels gradient difference and get the texture complexity of each image block.(2)For the RDH scheme with MHM,this paper redesigns the new global feature to replace the redundant local feature of the existing algorithm,and constructs the multiple prediction error histogram(PEH)by using k-means clustering.In the aspect of selecting embedding points,this paper proposes an improved crisscross optimization(ICSO)algorithm,which can precisely and effectively find the best combination of embedding points.The results indicate that the proposed method is superior than other similar schemes.Finally,this paper summarizes the related work of RDH technology,and describes the application scenarios of SHM and mhm technology respectively,and looks forward to the potential research direction of RDH algorithm in the future.