Research On Fast And Accurate Recognition And Positioning Of Vehicle Air Filling Port Based On Binocular Stereo Vision

At the moment,energy and environmental issues are receiving more and more attention from human beings.As energy sources such as coal and oil are increasingly depleted,and the environment causes greater pollution during use,the use of natural gas is becoming more and more important.More and more vehicles such as buses,trucks and private cars are also using natural gas as power source.However,most of the refueling work of gas filling stations is done manually by humans.Manual refueling is not only slow,but also has great safety risks.The development of technology,processes and equipment is imminent.For enterprise’s,reducing labor costs and improving the safety of aeration is an issue that needs to be solved urgently.This topic is to use this as an opportunity to design and develop an intelligent filling system.This article is mainly aimed at the filling conditions,a set of three-dimensional positioning system that can quickly and efficiently collect and process the filling port pictures,accurately identify the positioning of the filling port.The main research contents are as follows:(1)Design the image acquisition system.Select the appropriate binocular camera,light source,computer and other hardware equipment,and collect the gas port image and calibration plate image in the Visual Studio 2017(configured with OpenCV3.4.1)compilation environment to determine the gas port acquisition processing process.(2)Camera calibration and image correction.The camera calibration toolbox is used to perform camera calibration on MATLAB,and the parameters obtained by the camera calibration are used to correct the image.(3)Four-step image preprocessing.The four steps of image denoising,histogram equalization,histogram matching,and image sharpening are preprocessed on the left and right images.The noise and brightness differences in the image are eliminated,and the edge of the gas inlet is also enhanced.(4)Select the optimal feature point detection and matching method.Six new feature point detection and description algorithms were combined with three feature point detection methods and two descriptors,and experiments were performed on these six algorithms.Finally,the optimal algorithm combination of Fast feature points and SURF descriptors was selected.(5)Screen the feature points on the filling port.A set of criteria was proposed using the horizontal and vertical coordinates and radius of the center of the filling port edge to accurately select the feature points on the filling port.(6)Three-dimensional positioning of the filling port.The coordinates of the feature points on the filtered gas inlet are extracted,and the average value of the horizontal and vertical coordinates is used as the horizontal and vertical coordinate value of the gas inlet in the image.This standard value is substituted into the binocular disparity formula to obtain the three-dimensional coordinates of the gas inlet.10 sets of measurement data are selected for analysis.The average measurement error in the z-direction is 1.035%.