Research On Fuel Tank Cap Detection Algorithms Based On Image Morphology And Hough Transform

In the era of "mass entrepreneurship and innovation",unmanned intelligent systems such as UAVs,unmanned cars and unmanned supermarkets have sprung up after the rapid development of Internet +.Unmanned intelligent gas station is one of them.In the future,it will realize automatic identification of license plate,owner and oil products,automatic deduction of off-site money and automatic refueling link.Fuel tank cap detection is one of the key technologies of the unmanned intelligent gas station system,and has potential application value in the future intelligent charging system of electric vehicles.Therefore,it is of great significance to study the detection technology of fuel tank cap.In this paper,the detection of round fuel tank cap for small cars is studied.In this paper,the circular fuel tank cap of small car is taken as the experimental object.Aiming at the low detection rate of single Hough transform in fuel tank cap detection due to rainy weather and mirror imaging,this paper introduces image morphology and proposes a detection method based on image morphology operation and Hough transform,M&H algorithm.Specifically,Six kinds of morphological gradient operators are compared,and a combination of morphological gradient operator and Hough circle transformation algorithm is selected to apply to intelligent unmanned gas station.Experiments on Linux embedded platform show that the method effectively ameliorates the traditional algorithm vulnerable to rainy weather,mirror imaging and other interference problems,and improves the detection rate.Firstly,the theory of image processing required by the research institute is discussed,and the basic knowledge and mathematical tools of image,image scale transformation,image graying and image filtering are briefly introduced.The captured image is transformed by scale to meet the processing requirements.By graying,the irrelevant information of color image is reduced,the memory occupancy is reduced and the processing speed is improved.Two gray levels are needed in the experiment.The first grayscale is to transform the color image into three-channel grayscale image,which is for the convenience of color marking after the fuel tank cap is detected.The second grayscale is to transform the three-channel grayscale image into a single-channel grayscale image to prepare for Hough transform to detect the fuel tank cap.Through contrast experiments,the filtering method is determined,and whether it is filtered before the morphological gradient operator or after the gradient operator,so as to reduce the interference such as mirror imaging on the premise of guaranteeing the feature information of the target.Then the enhancement algorithm of fuel tank cap is analyzed and studied.The edge of fuel tank cap is highlighted by image morphological gradient operator to achieve the purpose of enhancing the target.Six kinds of gradient operators are obtained by analyzing image morphology expansion,corrosion,open and closed operations.Six image morphological gradient operators were compared by using the image data captured in the campus.The experiment shows that the expansion corrosion gradient operator has the best effect on the edge feature extraction of the tank cap.Finally,this paper takes the circular fuel tank cap of mini-car as the experimental object.The detection algorithm of circular fuel tank cap is analyzed and studied.The image morphological gradient operator is used to highlight the edge of the fuel tank cap,followed by Gauss filtering.Then Hough circle transformation is used to identify the fuel tank cap.Using 100 image data of fuel tank cap in different scenarios,experiments on Linux embedded platform show that M&H algorithm is superior to the single traditional Hough transform algorithm.