Multi-resolution image analysis based on singular value decomposition and subdivision

摘要

A new multi-resolution image analysis method based on singular value decomposition and subdivision is proposed in the paper. Subdivision scheme's good similarity among different subdivision layers makes the multi-resolution analysis has better approximation between the decomposed signals to the initial image. Non-symmetrical decomposition is employed in SVD. These properties provide the advantages of various sizes of transformation and more security. The experimental results demonstrated the effectiveness of the proposed multi-resolution image analysis method. Furthermore, extracted watermarks from tampered images were also clearly identified. The values of SNR are in the normal range..