Depth Information Extraction Based On Binocular Stereo Vision System

The information reflected by the object in Three-dimension Space not only includes height and width, but also depth information. The depth information plays an important role in the computer vision territory of research. Binocular stereo vision system is based the principle of person eyes observe the depth information. It use two cameras take the same picture of the calibration board and target object in different locations。At last, the feature point depth information is recover by the disparity and the location information of two cameras.Based on the studies of3D stereoscopic display methods, the active shutter stereoscopic display method is used in the paper.3D stereo images is generated by stitching images and watched by NVIDIA3D VISION Stereo Vision.A system of depth information extraction is proposed in this paper, which is based the research of binocular stereo vision theory. With the system, the camera calibration method is Zhang’s flexible new method, select feature point and get the circle centre point by Hough transformation, get matching double points with feature match, recover depth information by triangle measuring method. With the comparison of the recovery depth info and the measuring depth info, the experiment shows that the recovery method is reliable. With an object-oriented design methd in VC++, related algorithm software including camera calibration, feature extract, stereo matching and depth information extraction is developed.