Font Size: a A A

Research On Method Of 3D Reconstruction Of Under-vehicle Based On Binocular Linear CCD

Posted on:2015-04-19Degree:MasterType:Thesis
Country:ChinaCandidate:S N YuFull Text:PDF
GTID:2392330596979790Subject:Pattern Recognition and Intelligent Systems
Abstract/Summary:PDF Full Text Request
In recent years,the automobile accidents caused by entrainment of dangerous goods happened more and more frequently.Taking into account the difficulties of finding dangerous goods under vehicle chassis,this paper based on the sub module of "intelligent system of detecting unusual goods in vehicle chassis" to start the research.In this paper,the image data for three-dimensional reconstruction of vehicle chassis collected from bilnocular linear array CCD,the reconstruction based on linear array CCD including a linear array camera calibration,feature extraction,stereo matching,and 3D reconstruction.First we do a model of three-dimensional based on binocular linear array camera.Then we present a calibration method of binocular linear array camera based on self-calibration board after researching the coordinate system and the traditional plane array camera calibration methods.We resolve space relationship between two CCD into attitude angle of single camera and include angle of two cameras.We correct the imaging distortion of the two angles separately,experimental results show that the calibration results meet subsequent requirements.Secondly,in order to suppress the Influence of interference and dark light,we sharp the image edge and do de-noising preprocessing based on special structure of the smooth template.Then,we know under the array CCD imaging model,vehicle chassis has no rotation no zoom scale.According this feature,we proposed vertical canny edge as the features characterize of vehicle chassis.Finally,according to the features match criteria of the same content of binocular camera,we calculate the similarity of each left vision feature and right vision feature so as to give the similarity matrix.Based on the matrix,Get the final matches with piecewise recursive matching rules.With the right match results,we get the parallax of characteristic,then we can deduce the height message of characteristic,cording to the edge of the object and then get the full dense point cloud.Finally we use Opengl draw a three-dimensional model of the vehicle chassis.In this paper,we reconstruct dozens of vehicle chassis,and the experimental results verify the effectiveness of the three-dimensional reconstruction algorithm automobile chassis in this paper.
Keywords/Search Tags:binocular linear CCD camera, vehicle chassis, feature extraction, stereo matching, 3D reconstruction
PDF Full Text Request
Related items