The Research Of HDR Image Information Hiding Technology

High dynamic range(HDR)imaging techniques have received much attention in the recent years for abundant details and wide dynamic range of luminance.However,the HDR images display devices have high cost,immature technology and difficult to popularize in application.Therefore,in order to display HDR images and present abundant detail information on a common device,a tone mapping(TM)is required.However,the associated TM can be deemed as an inevitable attack in the field of HDR images copyright protection.Currently,information hiding technology is identified as one of effective approaches to resolve the problems of multimedia data covert communication and copyright protection.According to the characteristics of HDR images and the fact that TM is a kind of non-linear attack,this thesis studies the robust HDR images information hiding technology.The research contents are listed as follows:(1)In order to protect the copyright of HDR images,a robust watermarking algorithm based on redundant discrete wavelet transformation(RDWT)and singular value decomposition(SVD)for HDR images is proposed.Firstly,the RDWT and SVD are used to extract the structural information of the host image from the luminance component of HDR images,and serve as the carrier of watermarked.Then,combining the non-linear of TM attacks and the characteristics of human visual system,a hybrid mask is designed based on the luminance and texture to improve the robustness of the algorithm.Finally,for optimizing the imperceptibility of the algorithm,an adaptive selective mechanism of location is designed,and realized the embedding of watermark by adjusting the relationship between the selected pixels.The experiment results show that the proposed algorithm can effectively resist the attack of 20 typical TMs.In addition,it also has a super imperceptibility and higher embedding capacity in comparison with current algorithms.(2)In order to solve the problem that the traditional image processing methods destroy the internal structure of color images,a highly robustness HDR images watermarking algorithm based on tucker decomposition is proposed in this thesis.Firstly,tucker decomposition is carried out on the HDR color image to obtain the first feature map of the core tensor,which contains most of the energies in the HDR host image.And mark it as the embedding carrier.Then,the Auto-Regressive prediction method is used to establish a local correlation model on the first feature map,so that watermark can be adaptively embedded in the first feature map according to the prediction result.Furthermore,for optimizing the robustness of the algorithm,a low-complexity luminance masking approach is designed with modified specular free map.Finally,an embedding intensity selection strategy is proposed to balance the imperceptibility and robustness,and to adaptively select the embedding intensity for different HDR host images.The experimental results illustrate that the proposed algorithm is highly robustness against the attack of existing 27 TMs while keeping better image imperceptibility and embedding capacity.In addition,it has the obvious advantages in comparison with current algorithms.