Passive Ranging And Three-dimensional Positioning Based On Binocular Stereo Vision

Perception of 3D objects distance is one of the main tasks in computer vision. The scene's depth can be obtained through using camera to simulation the human eyes, which is the principle of binocular stereo vision technology. It has the advantages of fast, non-contact measurement, simple and reliable, and is widely used in civil and military aspects. It is also already a hot research topic in the field of computer vision.Typically, a complete binocular stereo vision system includes image acquisition, camera calibration, image pre-processing and feature extraction, stereo matching and three-dimensional reconstruction. Currently, 3D measurement based on binocular stereo vision has conflicts between accuracy and efficiency. In order to solve this problem, this article makes some improvements on the camera calibration, feature extraction and matching, allowing the binocular vision system to achieve high accuracy and fast measurements. Details are as follows:Firstly, the 3D measurement principle based on binocular stereo vision and error analyses are described. Secondly, for binocular camera calibration, zhengyou Zhang calibration method is chosen and hole is selected for calibration reference. The center of the hole can reach sub-pixel accuracy through ellipse fitting on the edge of hole, which can greatly improve the accuracy of calibration. Experimental results show that the calibration accuracy of two cameras is relatively high. Thirdly, According to the fast measurement of demand, an improved method based on FAST and Harris is proposed, experimental results show that the detection algorithm is fast and effective. In terms of point matching, using cross-matching with BRIEF and epipolar geometry constraints to reduce mismatch greatly. Finally, the 3D coordinates of the point in the world-space can be obtained with the results of camera calibration and the coordinates of matched points. Experiments show that the binocular stereo vision system can reach high ranging precision in a certain range.For passive ranging and three-dimensional positioning based on binocular stereo vision, the method implemented in this paper can measure quickly while maintaining a high accuracy in a certain range.