Research On Binocular Vision Positioning Algorithm For Small Municipal Public Facilities

With the construction and development of smart cities,The management method of municipal facilities should also be developed in the direction of intelligence.At present,the management of municipal facilities adopts the traditional method of manual inspection,which can no longer meet the needs of urban development.In response to the above problems,this article proposes a binocular vision positioning algorithm for small municipal public facilities under the background of binocular stereo vision technology.The main work of the article is as follows.First,the mathematical realization process of binocular 3D vision ranging is analyzed step by step and the main principle of camera imaging is deeply analyzed.The key parameters of the binocular camera are corrected by MATLAB.The corrected camera can effectively avoid the influence of lens distortion on the final result.The calibrated binocular camera is used to shoot the checkerboard.At this time,combined with Zhang Zheng you calibration method,the checkerboard views in different directions are captured,and finally a binocular vision picture that meets the requirements is obtained.Secondly,after gray-scale processing and filtering to denoise the binocular image,feature extraction is performed on the detected target.In order to solve the problem of uneven distribution of feature points extracted by existing algorithms,a uniform extraction algorithm for image feature points is proposed.According to the current mature feature point extraction algorithm ORB algorithm,the feature points extracted in the image are divided and screened in a quadtree,and the algorithm steps of feature point extraction are re-introduced.The experimental results show that the algorithm has obvious effect on the uniform extraction of feature points on the basis of the original algorithm.Finally,according to the principle and implementation of the stereo matching technology,the local and semi-global matching algorithms are studied.Through the analysis and summary of the stereo matching algorithm,the use of the semi-global algorithm can achieve a better matching effect.Therefore,the SGBM(semi-global)algorithm is used to process the pictures taken by the calibrated binocular camera.The algorithm can output the left and right disparity information of the matching point,which can be used to calculate the target depth.At the same time,the stereo matching algorithm experiment is performed to verify the accuracy of the algorithm.Under real road conditions,the pavement target is spatially located.After the experiment,the experimental results are compared and analyzed.The experiment shows that the algorithm proposed in this paper has a positioning error of less than 5% for the spatial target,and can realize the localization of small municipal public facilities.