Research Of Eccentric Bucket Filler Mouth Vision Positioning Key Technology

Using eccentric bucket to store and transport liquid is a relatively common phenomenon. Like oil and chemical raw materials, even the food industry has also used eccentric bucket more or less. For filling liquid itself, which had different varieties, so the selection of filling mode also is different, which including some toxic liquid with a penetration ability, like HF and NaH and so on. When filled, it is unavoidable for workers who will direct contact with this liquid, so the caused damage is inevitable. But whether a new filling way can be designed, which can realize automatic production in factory, and no need of worker directly involved in production, while intelligently remote controlling basically can reduce or avoid the harm of toxic liquid for workers. In the aspect of filling difficulty, as filler was eccentric, result in the eccentric barrel which move to the filling station randomly spread on the circumference, so the mechanical hand can not repeat the same process to finish filling. It is necessary to apply visual technology to the filling system. Firstly achieving identification and position for filler, and transporting the location information to the computer for processing and analysis, and then complete the real-time controlling of the manipulator, lastly achieving filling. As can be seen, the key technology is automatic positioning of filler.In China, filling machines have different varieties, where function and realization way has be different, with the research from the subject group, a new automatic positioning system is developed to achieving the recognition of eccentric bucket filler. The system is based on the theory of computer vision, and giving full play to the advantages from the vision system, not only reducing the labor intensity of workers, but also avoid the harm of toxic liquid and permeate to the body.The image edge acquisition of eccentric bucket filling mouth and fitting of the two numerical image parameters is the key content of this thesis. Firstly this paper introduces the detection system hardware structure and software flow chart, and use the means of Zhang Zhengyou to camera calibration, achieve distortion correction for camera parameters, analysis the main causes of error; And then select Gauss filter to complete the image smoothing, under the premise of guarantee target contour edge in variant fuzzy, to get rid of well the disturb of noise on the identification process; For image segmentation, it is improved based on the fixed threshold segmentation, to develop a dynamic threshold segmentation method, realizes the automation selection of threshold. In order to solve the problem of the image gray value change which is caused by the position of different images is different and the change of light condition. Achieving the edge detection by using the Sobel operator. And in order to speed up the recognition process of the system, the paper introduced the NCC template matching algorithm, meanwhile the algorithm is improved, combined with the pyramid layered search strategy, completed the rapid positioning of eccentric bucket filler; Finally, using circle fitting algorithm based on robustness to achieve the fitting of the filler and the recognition of the center.After verification by experiment, the construction of the system meet the demand which the fast automatic positioning of the eccentric bucket filling hole in the automatic filling process. And the positioning accuracy of automatic positioning system can reach 0.085～0.090mm.