An Object Reconstruction Method Based on Binocular Stereo Vision

This paper presents an object reconstruction method based on a binocular stereo vision system. First, the relative position and orientation between the two cameras of the system are obtained by a binocular calibration process. Then feature pixels are extracted from images of the two cameras of the same scene according to the SIFT (Scale Invariant Feature Transform) feature. Feature pixels on the image of one camera are matched with those of the other camera according to differences between their SIFT features. And the RANSAC (Random Sample Consensus) algorithm is used to eliminate incorrect matched pixels. Then a 3D coordinate point is obtained from each pair of matched pixels. Finally, 3D models are constructed from the 3D coordinate points through triangulation and texture mapping. In the above processes, a uniform method of calculating coordinates of 3D points from pixel pairs is introduced, which is fitted for arbitrarily orientated optical axes of the left and the right cameras. Experiment results show that the proposed method can obtain 3D points sampled from real objects and produce 3D models consistent with reality.