Research On Oil Pumping Pipe Count Image Recognition Algorithm And Simulation Analysis

Oil is the main energy and it has an important impact on industrial production and daily life. Factory will use a large number of oil pumping pipes in pumping oil, the count of oil pumping pipe is essential management needs. Now not only the efficiency of artificial count is bad, but also the accuracy of count is low.This paper use image recognition technology to automatic count of oil pumping pipe by collecting end image of oil pumping pipes. In practice, not only to improve the efficiency of preparation work, but also other type of pipe can be used in automatic counting. In the industry, it has great value and economic benefits. The main contents of this paper include:oil pumping pipe image mosaic, oil pumping pipe image preprocessing, edge detection of oil pumping pipe, image recognition algorithm of oil pumping pipe, oil pumping pipe count and experimental analysis.When image acquisition, because of the limited distance of shooting and reduce edge of image distortion, we use a single CCD can not meet the application requirements. We need to use multiple CCD splice together to collected images, then we use image mosaic method to splice pumping pipe images that there is much overlap. The paper uses image mosaic method that will put the image projected onto the cylinder, that uses the phase correlation method to find matching points, that finally uses the trigonometric transition smooth method to fusion images.Oil pumping pipe image preprocessing, including image format change, the change of gray image, image binarization, image enhancement, image edge detection. Avariety of methods are used in each processing. Through tested and analyzed, we are select suitable method to process oil pumping pipe images.This article has used the classical edge detection methods and the method which use detection of the morphology and boundary tracking to combining to detect image border of oil pumping pipe. According to the experimental analysis, we will use based on morphological edge detection and edge tracking combined algorithm, which can provide good boundary, in this paper.In recognition of the image, as the oil pumping pipes border after shooting the image actually appears elliptical,this paper proposes an improved method of circle and ellipse Hough transform. It will use the geometric properties of elliptic and circular to reduce the Hough transform dimension and can reduce the amount of calculation and statistics. We use the law to take specific point method to reduce the number of invalid access point,to reduce the amount of computation, and to improve recognition speed.Finally, we are count of oil pumping pipe Then, for image of neat oil pumping pipe, we use the standard pipe size to match detection of the size, output statistics. For ultra-wide indent oil pumping pipes image,we needs amend. This paper made a limited theory to amend indented elliptic under certain conditions.