Feature Space Based Watermarking in Multi-ImagesOutlinesIntroductionIntroduction(2)Introduction(3)Feature Based WatermarkingEigen-Decomposition and SVDEmbedding One Bit (1)Embedding One Bit(2)DetectionEmbedding in Multi-ImagesSimulation Results (1)Simulation Results (2)ConclusionFeature Space Based Watermarking in Multi-ImagesXin ZhouOutlinesIntroductionFeature Space Based WatermarkingSimulation ResultsConclusionIntroductionGoal:Implement the watermarking in a set of images or video based on eigen-decomposition or SVD (Singular Value Decomposition)Introduction(2)Where to embed bits?Spatial Domain: Directly process original values of the host image according to the watermarkFrequency Domain: Transfer the host image into another domain and change the corresponding coefficients according to the watermarkIs it possible to embed bits in the feature domain?Introduction(3)Other’s WorkUsing SVD (Singular Value Decomposition) to embed bits in one imageUsing ICA (Independent Component Analysis) to detect the watermarkWhat I want to doFind a method to use eigen-decomposition or SVD to embed watermark in multi-images or videoIt will be more robust to embed bits in each imagesMore difficult for others to estimate the watermarkFeature Based Watermarking Eigen-Decomposition and SVDEmbed one bit DetectionEmbed watermark in a set of imagesEigen-Decomposition and SVD mxn matrix A, of rank r, can be expressed as the product: A = U * S * VTU is mxr term matrix S is rxr diagonal matrixV is rxn document matrixIf A is nxn matrix, r=n, we have A *U= U * SEmbedding One Bit (1)Assume the host image is a mxn matrixPerform the SVD to get S matrixEmbed one bit in the S matrix according towhere {si}: original coefficients{si’}: marked coefficients{b}: the bit to be embedded which is 0 or 1k : watermark strength, adjusted by the just-noticeable- difference (JND) standard'i is s k b= + �Embedding One Bit(2)After Embedding:A’ = U * S’ * VTwhere S’ is the watermarked singular matrixA’ is the corresponding watermarked imageDetectionAssume we get the watermarked image A’.Perform eigen-decomposition or SVD to get the S’Compare S and S’, we can get the watermark'i is sbk-=Embedding in Multi-ImagesMethod I:Generate a pseudo random codebookDecide which bit should be embedded to which image based on the codebookFor a specific image, use the previous method to embed bitsMethod II: Use QIM-like method to decide which bit should be embedded to which image.Simulation Results (1)Simulation Results (2)ConclusionUsing eigen-decomposition or SVD to embed watermark into multi-images. Implemented basic functions of the proposed methodNeed to do more tests under various
View Full Document
Unlocking...