Delaunay triangulation mesh-based motion estimation and compensation based on redundant discrete wavelet transformation and improved SIFT algorithm
Liang-Hui Wang · Jisuanji yingyong yanjiu · 2011
To ulterlorly improve the effect of motion estimation and compensation,this paper proposed a delaunay triangulation mesh-based motion estimation algorithm based on redundant discrete wavelet transformation(RDWT) and improved SIFT algorithm.First,it extracted the potential motion area(PMA) and designed a adaptive template for feature points extracting in RDWT domain.Then proposed an improved SIFT algorithm.It could extract and match the feature points fast and accurately and generate DT mesh.Finally,a delaunay mesh-based motion estimation and compensation was done in PMA by the affine transform.The result of experiment indicates that this method can quickly and efficiently extract stable feature points of video image.It can maintain a higher peak signal to noise ratio(PSNR) and improve motion estimation efficiency,and the subjective quality of image is good.This approach has some performance in PSNR and efficiency compared with the existing irregular triangular mesh-based motion estimation algorithm.