Research And Realization Of Surface Reconstruction Algorithm Based On Feature Points Matching

Binocular stereo vision mainly studies how to form an image of the same scene with two cameras taken from different locations based on geometric principles to achieve the recovery of the three-dimensional scene. Recently, with the rapid development of3D data acquisition technology, surface reconstruction and feature point matching technology has gradually become a hot research field of computer vision, and it is widely used in intelligent robots, production automation, medical and many other areas. So studying the three-dimensional reconstruction is of important theoretical significance and practical value.On the basis of the summarizing and analysis of materials about surface reconstruction and feature point matching from home and abroad, This article provides an overview of the feature point matching, surface reconstruction, and camera calibration computer vision technology, and studies some problems of surface reconstruction and feature point matching, discusses the basic content of feature point matching technology, and study intensively the typical feature point matching algorithm----SIFT (Scale-invariant feature transform).This article makes a summary evaluation of SIFT and points the strengths and weaknesses, using specialized corner matching method, proposes the corner points matching algorithm based on SIFT. A description of the image matrix is created using the new processing method, inherent matrix constituting the image is found according to the information of the fundamental matrix, and projection matrix is solved the binding matrix deformation theory, and finally adjusting the specific pixels, approximate shape is sketched out to achieve the object surface reconstruction.