The Research Of Stereo Vision Reconstruction System Base On Structured Light

Stereo vision is a branch of machine vision.As technology continues to mature it have wide apply in industrial robotics, navigation, automated inspection, monitoring and other industries in recent years. It simulate the human eye, using two cameras capture the image from different spatial angles, and calibration camera to obtain the corresponding spatial relationship between the camera through the scene, the scene is obtained the three-dimensional information from two images.This paper show how to create the stereo vision based on the structure lighting. The structure lighting has high-precision and high-efficiency characteristics. Using structure lighting creates stereo vision system. This paper’s contents show as blow.(1) The researching on camera linear model.Introduction to linear model of camera, analyzed camera intrinsic matrix parameters and introduction the meaning of the distortion parameters, and the lens horizontal distortion and tangential distortion, and the chess board as the target, obtain the camera intrinsic parameters and distortion parameters.(2) The researching on stereo imaging model.Based on linear model we analyzed geometric model of stereo vision system, in accordance with the polar constraint in the model, polar constrained search range from2D plane can be narrowed into a one-dimensional line, in order to improve computational efficiency and matching efficiency. Analyze the relationship between the basis matrix and polar constraints.(3) The researching on stereo rectifies.Introduces the geometric model of stereovision, introducing the polar constraint, establish a polar constraints, respectively, the use of uncalibration rectify and calibration rectify:BOUGUET algorithm, derived the process of transformation. Rectifying the left and right images, eventually get the line alignment and distortion less images.(4) The researching on stereo matching.According to rectified image by stereo calibration, looking for space point in the image of the corresponding matching points, introduces various matching methods, and based on the constraint conditions met measurement function, compare various stereo matching strategy. And it is implemented in a c++program.(5) Proposed stereo matching strategy based on ECA algorithmsIs put forward based on the ECA algorithm of optimal search strategy to the line on the image to the stereo matching, puts forward to SAD, NCC, ZNCC as evaluation function, constructing polar, uniqueness constraints, sequence constraint rules, compared to traditional search algorithm to optimize search strategy and ECA strategy under the SAD, NCC and ZNCC evaluation function algorithm, the effect of matching and the matching of time under the running figure iteration of the algorithm.(6) The researching on the center of the structure of the light extraction methodsArticle compared the traditional optical center extraction method, based on the consideration of the light stripe shape and boundary information, this paper proposes a rapid extraction method, the light stripe and comparing and traditional method, the experimental results show that the method has strong usability and has quicker extraction rate.(7) The researching on stereo vision3D reconstruction.To get parallax graph based stereo matching, depth information feature points extraction light bar.According to feature point set, the Delaunay triangulation subdivision algorithm analysis, on the basis of analysis based on depth image3D reconstruction process. And build the MFC based on OpenGL development platform, to the standard image and the experiment of images taken on stereo vision system, the space and the depth of the feature point information extraction based on OpenGL3D scene reconstruction, and has carried on the experiment and result analysis.