A novel method of DT mesh motion estimation and compensation based on RDWT
Jianxun Wang, Ming Zhen Yu, Ying Xia, Jichuan Chen, Zheng You Xia · 2012
In order to make further improvement for the effectiveness on motion estimation and its compensation of mesh, this paper proposes an algorithm with mesh motion estimation of Delaunay Triangle (DT), based on redundant wavelet transform (RDWT). Firstly, the feature points and potential motion areas (PMA) of mesh are extracted from the domain of RDWT, and the motion intensity degree of image blocks can be accordingly classified. Subsequently, the different searching strategies, on the basis of the above-mentioned situation, are adopted for the image blocks of various motion types, thus, the quadrilateral searching range can be put forward. Moreover, the adoption for motion estimation of mesh vertex will be integrated with the block-matching method of motion estimation and its compensation within time-based domain of the potential motion area, while the motion compensation can be accomplished by means of triangle affine transform. It can be shown from experiment that the desirable results have been achieved with this algorithm in the aspects of PSNR and compensatory effectiveness.