This paper presents a stereo matching algorithm which combines the genetic algorithm with the disparity propagation scheme to generate the dense disparity map. We design the detailed steps of genetic algorithm to realize the main stereo matching process. In order to increase the efficiency of this algorithm and improve the quality of the disparity map, we only perform it on the lower resolution image after the pyramid division of the original image. Then the uniformity distribution model performs as the disparity propagation stratagem to generate the final disparity map. We tested the proposed algorithm on the stereo images, and the quality of the disparity map demonstrates the effectiveness of it. Keywords-stereo matching; genetic algorithm; propagation stratagem; disparity mapI.978-1-4244-3894-5/09/$25.00 ©2009 IEEE

Image retrieval is a hot topic in the field of computer vision. Displaying query image and retrieved image as three-dimensional images can facilitate the users to evaluate the quality of the retrieved results. To show a single image in three-dimensional appearance, we propose to generate stereo image based on both physiological and psychological stereoscopy. We have found out how psychological stereoscopy can strengthen or weaken physiological stereoscopy. On the basis of this, we propose to introduce random binocular disparities into the single image. The net effect of this operation is the better stereo perception. When compared with other algorithms, the proposed algorithm is simple and the computational cost is relatively low. The effectiveness of our algorithm is demonstrated on different types of images. Since the 3D images are difficult to be captured, the algorithms we proposed aiming to generate stereo image from a single image are of great importance to the development of the field of the 3D vision.

No abstract

No abstract

No abstract