Research Of Machine Vision Based Position Method For Automatic Cement Filling System

Most Chinese cement enterprises use cement filling vehicle for cement transportation, and cement filling efficiency has great influence on enterprise benefits. At the moment, most enterprises join charge hole and discharge hole by hand during cement filling. Manual operation has several shortcomings such as large labor intensity, slow speed and easy occurrence of accidents, so it is of great application value to develop automatic cement filling system. As an important part of cement automatic filling system, this dissertation researches on the automatic butt-joint method of charge hole and discharge hole based on machine vision technology. Its main work is as the following:(1) Positioning schedule of eye-to-hand and eye-in-hand combination:Automatic cement filling has requirement of wide field of vision and high positioning accuracy. Therefore, this dissertation presents positioning strategy of eye-to-hand and eye-in-hand combination. First, eye-to-hand camera is used for gross position in large scale, and then eye-in-hand camera is used for further accurate positioning.(2) Pixel level precision extraction of charge hole in strong interfered picture:It is the basis for the realization of the automatic filling to extract charge hole accurately.. Because of many interference factors such as much dust in air and unstable sunlight, the image is of poor quality, which makes it an challenge to extract accurate charge hole. The method first take use of morphological closing operation for edge enhancement; then median filtering based Canny operator is used for edge extraction; Last, it adopts continuous isometric string based fast circle detection algorithm to extract of charge hole.(3) Iterative approximation based Precise positioning method of charge hole and discharge hole:Due to the influence of movement error of mechanical arms, it is difficult to realize accurate positioning of charge hole and discharge hole in one time. This dissertation uses iterative approximation method to realize accurate positioning. First, it adopts direct linear calibration method to calculate coordinate transformation parameters according to imaging model of the visual system; Then it calculates mechanical arms movement parameters according to the position of outlet in the image and conversion parameter. Get new images again after the movement of mechanical arms to judge whether it has accurate positioning already. If not, detect the charge hole and calculate movement parameters of mechanical arms repeatedly.