A Novel Stereo Image Coding Algorithm Based on Delaunay Triangulation Mesh

摘要

Stereo image compression is of growing interest because of new display technologies and the needs of telepresence systems, A variety of algorithms have appeared in the literature that make use of the framework of block-based disparity estimation and disparity-compensated residual coding, which is similar to the monoscopic image framework. So that it has several drawbacks, for example, blocking artifacts. In this paper, a novel adaptive DT mesh based stereoscopic image coding algorithm is proposed. The algorithm makes use of adaptive DT mesh to estimate disparity vectors, and EZDCT based residual image coding. The algorithm is performed in three steps: adaptive DT description for the reference image, disparity vector estimation based mesh and simple occluded regions detection, and disparity vector and residual image coding. Much of this work concentrates on improving the veracity of disparity estimation using DT mesh model and encoding the residual image appropriately. Simulation results show that the proposed scheme achieves higher veracity of disparity estimation at least 10%, as compared to conventional algorithms. And compared with conventional compression algorithms, the algorithm improves average PSNR about 1 db. Especially at low bit rates, this algorithm achieves better perceptual quality of decoded image, as compared to a block-based coding scheme.