Research On The Approach Of Feature Matching And Novel Image Synthesis

Features of image denote that the change of gray value is obvious for points, lines or areas. Feature matching involves establishing corresponding relationship for features which are extracted from reference images between images based on the theory and method of stereo matching. The approach of image synthesis generates images for novel viewpoints by use of feature correspondence between reference images. The approach of feature matching and image synthesis are important techniques in the image-based method to computer graphics.Stereo matching is a important task for the study of computer vision, a stereo matching algorithm based on corner detection is studied in this article. The smallest univalue segment assimilating nucleus (SUSAN) approach is used to decide whether the pixels on the edges to the left and right images are corners, and the initial correspondence relationship by the retained corners is established between the corresponding corners of the left and right images according to the similarity of the features. Then the fundamental matrix could be computed. The initial matching result is iteratively improved by using the fundamental matrix to discard the false matching corners, then the ultimate matching result is obtained and the fundamental matrix is accurately calculated. The matching relationship between the pixels on the corresponding epipolar lines of the left and right images is searched using dynamic programming approach. Therefore, the corresponding relationship of all the pixels between the left and the right images can be established.Image Morphing is such a technique that could build a series of intermediate frames between two reference images in which content and color may be quite different. The algorithms based on feature line pairs are studied in this article: An image morphing algorithm based on feature lines grouped is developed as to the problem of occluded image morphing between two reference images of the same scene with different viewpoint according to the disparity variety of the feature points on the foreground and background images, then the morphing operation is respectively processed in terms of each group of line pairs; Multiresulztion image morphing approach could be used to generate several sequences of intermediate frames with variety of details transition, the alterable rectangle areas with sampling points centered are sampled to acquire the colors of desired pixels. The view morphing approach based on the feature lines uses three-step algorithm, prewarp-morphing-postwarp, tomaintain the contour features of objects on the reference images as the intermediate frames are built.The technique of image-based 3D reconstruction by means of images of the same scene with different viewpoints is studied. Camera self-calibration and 3D reconstruction for corresponding feature points are key problems in the 3D reconstruction technique. These problems are researched in this article. Numeric simulation computing is processed and a better result is obtained.