Research on Camera Localization Based on Monocular Vision
|Course||Vehicle Operation Engineering|
|Keywords||Camera positioning Camera Calibration Monocular vision ranging Image Stitching Image registration Image Fusion|
Computer vision because it contains a wealth of environmental information has been widespread concern , one of the monocular vision - based positioning technology is the computer vision research direction . General steps based on monocular vision camera positioning method first pass between the image and the template matching relationship corresponding points to create a constraint homography , then homography solving the camera position parameters . However , the corresponding points between images , and the corresponding points between the image and the template matching has been the difficulty of the image processing and analysis , the matching correctness directly affect the positioning accuracy . Therefore, the positioning process to minimize or avoid matching is very practical value . In view of the above problems , this paper two camera positioning : positioning method ( triangle method ) based on the geometric relationship between the camera and the camera positioning method based on image stitching . Ranging monocular vision model based on the geometric relationship between the camera positioning method , establish the geometric relationship between the camera and the scene known feature points can be derived through a series of triangular camera dimensional coordinate plane position , the method to avoid the positioning process on the image point matching problem , the continuous positioning of the camera can be achieved . Camera positioning method based on the image stitching and its essence is use of the affine transformation of the image splicing process , i.e. after the plurality of pairs of matching points in the two adjacent positions in the image is found , solving the affine transformation model , to thereby obtain two images transformation matrix and translation matrix , and launch the camera position changes , the initial camera positioning . At the same time, the image of the scene to get a larger field of view image stitching , more environmental information . The experiments show that the two positioning methods of this study is feasible , there is a certain degree of robustness , initially completed a monocular vision camera positioning method based exploration and research , for the realization of the lunar rover intelligent vehicle based visual track speculate and tracking provide the necessary technical support .