The Reconstruction, Three-dimensional Information Of The Moving Target Based On The Complexity Of The Context Of Binocular Vision

The 3-D reconstruction technology of computer vision is a significant field of computer vision,and binocular stereo vision is a very important technique in 3-D reconstruction of scene.When use it,we first get two pictures respectively by left and right cameras, then wo could get the 3-D information included in these pictures by geometry and other relationship.To build a experimental platform of binocular stereo vision as a starting point,this paper deeply discussed how to get the precise 3-D linformation of moving target center of scene under complex background,in the binocular stereovision system. This paper has studied mainly on four questions:binocular stereo vision camera calibration, binocular stereo vision-based moving target detection, area correlation-based stereo matching algorithm and feature-based stereo matching algorithm, MFC framework-based binocular stereo vision software development platform.On binocular stereo vision camera calibration:first use chess board as calibration object, get the corner coordinates information through matlab simulation,and use Sub-pixel approach to improve the coordinates accuracy.Then use Plane pattern two-step method to solute camera parameters,which include camera internal parameters, camera external parameters and camera distortion parameters. Experiment results show that the algorithm is effective.On binocular stereo vision-based moving target detection:based on the background subtraction algorithm under static background of monocular vision,this paper proposes an objects detection method combining gray and background subtraction algorithm under complex background based on binocular vision. Experiment results show that the validity and accuracy of this moving target detection algorithm for the complex background (with noise and shadow interference).On stereo matching and 3-D reconstruction:In practice, scenarios are often subject to light changes and noise,and in most occasions,people only focus on the 3-D information of moving target in the scene.First this paper get disparity map by using area correlation-based stereo matching algorithm,then combining background subtraction algorithm to accuracily detect the moving target under complex background. Then use feature-based stereo matching algorithm on he moving target result for Secondary stereo matching.Last get the precise 3-D linformation of moving target with the result of binocular stereo vision calibration. Experiment results show that the algorithm is robust.On binocular stereo vision software development platform:first get the left and right video information through the directshow platform provided by microsoft company.Then call on matlab in the MFC framework through the matlab engine.After processing,use setbitmap member function provide by static class to show result. By system demonstration, show that the platform can make research more focused algorithm itself.