Font Size: a A A

The Conicoid3D Reconstruction Based On Binocular Vision

Posted on:2014-11-05Degree:MasterType:Thesis
Country:ChinaCandidate:H W LiFull Text:PDF
GTID:2268330422453239Subject:Signal and Information Processing
Abstract/Summary:PDF Full Text Request
Binocular stereo vision, one of the hotspots in the field of computer vision, hasimportant theoretical significance and practical value. The basic principle of binocularstereo vision mimic human eyes’ visual perception process, observe the same objectfrom two different viewpoints and restore the image’s three-dimensional informationthrough triangulation method. It has made some achievements that to achieve thestructure reconstruction of the three-dimensional space object according to theinformation of the two-dimensional image. However, commonly reconstruction methoduses polyhedron only with plane and straight line as the reconstruction object, which hasa lot of limitations; the reconstruction of natural or man-made curved objects still facessevere challenges.Conicoid is a commonly used model of industrial objects; the three-dimensionalstructure reconstruction of conicoid is one of the basic problems in the field of computervision research. In this paper, a binocular camera is used to obtain two-dimensionalimages of the same scene from different angles, and conicoid contour’s reconstructionmethod is researched. Mainly from the camera calibration, image edge contourextraction, contour curve matching, structural reconstruction carried out a detailedelaboration and implementation of technology research, to obtain the contourinformation of three-dimensional space from the two-dimensional image scene’scontour.Firstly, the basic principles and methods of camera calibration are introduced,detailing Zhang Zhengyou calibration method and the experimental results are given;and then an improved edge detection method based on canny operator is proposed toextract the image edge contour through the characteristics of the image scene analysis.Using curvature fractal-based approach cut the contour curves into different curvesegments when matching the contour curve. The curve segment matching method isgiven based on the detected feature point on the curve through analyze the geometricrelationship between points and line segments.In terms of the contour reconstruction, according to the nature of the spacequadratic curve, combined with the existing reconstruction methods of space points and spline curves, for calibration image, curve segments as primitive of three-dimensionalreconstruction, a new method for the three-dimensional structure reconstruction ofspace quadratic curve is proposed and applied to the contour reconstruction of theconicoid.
Keywords/Search Tags:Computer Vision, Contour, Conicoid, 3D Reconstruction
PDF Full Text Request
Related items