To combat the problems of disparity image inpainting, an improved exemplar-based image inpainting method is proposed. The original stereo image we select which corresponds to disparity image is decomposed by TV-based decomposition and the structural information is obtained. The patch priority in the position corresponding to the disparity image is calculated by using the structural information, which reflects true characteristics of the original stereo image and disparity image. Experimental results show that inpainting performance is better with the guidance of this patch priority for disparity image.

The stereo matching algorithm based on aligning genomic sequence is proposed in the paper. This method is divided in three steps: do genomic sequence on the same name epipolar of stereo matching, get branch matrix by establishing genomic sequences using the same name epipolar after being genomic sequences, control points technology and dynamic backdate method to get disparity. The experimental results show that stereo matching method based on genomic sequences has fast speed and good matching quality.

No abstract

No abstract

No abstract