Object-based coding of stereo image sequences using joint 3D motion/disparity segmentation

An object-based coding scheme is proposed for the coding of the right channel of a stereoscopic image sequence, using motion and disparity information. A multiresolution block-based motion estimation approach is used for initialization, while disparity estimation is performed using a pixel-based hierarchical dynamic programming algorithm. A split and merge segmentation procedure based on both 3D motion and disparity is then used to determine regions with similar motion and depth parameters. This is combined with an efficient depth modeling method that offers full depth information at the decoder site. In order to reduce the computational load of the merge phase of the algorithm, a fast algorithm is implemented which speeds up the merge procedure considerably. The segmentation part of the algorithm is interleaved with the estimation part in order to optimize the coding performance of the segmentation procedure. Motion and depth model parameters are then quantized and transmitted to the decoder along with the segmentation information. An object-based motion compensating scheme is then used to reconstruct the original image, based on the objects created by the segmentation approach.