The Applied Study Of Binocular Stereo Vision In Reverse Engineering

To modern enterprises, making use of reverse engineering is an efficient method of technique innovation, updating products and satisfying the fluctuant market requirement. It's a key problem for reverse engineering how to acquire the models of original objects and process them for reconstructing 3D models, mutual design and innovative design, then produce new products and put them into the market.The stereo vision is one of the important methods to reconstruct 3D models of hypostatic object, and in this paper, in order to apply the binocular stereo vision in reverse engineering, the experimental system built to reconstruct the model of free-form surfaces, and the camera calibration and image matching of stereo system are deeply researched.According as the basal principle of binocular stereo vision, the binocular stereo vision experiment system based on two CMOS is built. Then, the facility and stability of Zhang's calibration method are ascertained after analyzing several camera calibration techniques. Based on Zhang's method, the calibration method based on the centers of circles on a 2D plane is put forward. By contrasting this method with another two methods, based on corners of squares and based on centers of squares, the validity of this method presented in this paper is proved. The implicit camera system model is made with BP neural network simulating the relationship between the 3D objects and its images in the stereo system, which avoiding the system errors caused by unperfected mathematic relation. Experiments show that the high calibration precision could be got by this way.In this paper, the image processing methods, Gauss smoothing and filtering, Canny edge detection and Harris corner detection are used after contrasting several image pre-processing techniques. Then, the image matching technique based on Scale Invariant Feature Transform is studied, and the fake matching points are eliminated by adding epipolar restriction in this method. A new matching method which combines feature matching and area matching, edge feature and corner feature, is depicted, and the experiments show that the matching time could be reduced, and precision is improved by using this method. Synthetically the techniques dissertated above, the experiment on get the 3D information of an object is done, and the technique of reconstruct 3D models of free-form surfaces in reverse engineering is discussed at last.