Research On 3D Reconstruction Algorithm Based On Binocular Stereo Vision

Binocular stereo vision is a complex subject and it involves complex application of multiple disciplines which includes image processing, computer graphics, signal processing and mathematics. Three-dimensional reconstruction is the core part of binocular stereo vision and it is used frequently in robot navigation and visual surveillance and in the manufacturing sector also has its presence. With the rapid development of computer virtual reality technology in recent years, the field of three-dimensional reconstruction applications has extended to the technical aspects of drawing of image,3D animation and so on.To get an accurate three-dimensional image reconstruction algorithm, firstly this passage analyses the classic three-dimensional reconstruction algorithm through the study of camera calibration, image preprocessing, feature extraction, stereo matching and the display of three-dimensional reconstruction and draw a three-dimensional reconstruction algorithm of higher precision and faster speed. In terms of camera calibration, it is a calibration method which is based on Zhang Zheng You flat calibration method and camera of nonlinear models, this calibration method operation is simple, has the feasibility of high, and the high accuracy of the calibration results can be obtained. In the image preprocessing, this paper describes the principle of the three algorithms that median filtering, histogram equalization, Laplace sharpening. And it preprocess the image by median filtering, histogram equalization, Laplace sharpening respectively. In the feature point extraction, this paper analysis the Harris algorithm, subsequent it extracts corner by using Harris algorithm. In stereo matching, correlation matching method, RANSAC method and the minimum value method are analyzed and compared, then this paper proposes an improved stereo matching algorithm, that realizes matched with high accuracy to corner extraction by using RANSAC method and Zhang Zheng You matching algorithm, matching accuracy can reach above 90%. Finally, applying the principle of parallax to calculate the three-dimensional coordinates of matching points and using VC++and OpenCV programming algorithm on the basis of triangulation algorithm, and then get a accurate three-dimensional reconstruction of the target object diagram. And the three-dimensional reconstruction algorithm in this paper is applied to the different characteristics of A and B two scenarios, also get a accurate three-dimensional reconstruction renderings, test fully the applicability of the algorithm.