Research And Application Of Video Detection And Recognition Method For Coal Flow At Coal Unloading Point Of Mine Hoist

The mine hoist is an important equipment for coal transportation.With the continuous development of intelligent mine,more and more intelligent devices are applied to the mine to assist the mine staffs to identify and discriminate abnormal production.With the improvement of coal mining capacity,the mine hoist needs to load more and more coal.Sometimes,the sector doors of the mine hoist are not opened or closed abnormally,or sometimes there is too much coal that causes the grate to be blocked.However,there is currently no effective method to detect the above-mentioned problems.Therefore,this article uses image processing and pattern recognition technology to study the state of the sector doors and the amount of coal flow,and the effectiveness of the method in this article is verified through examples analysis.The research contents of this thesis are as follows.Firstly,the camera collects the field images of the sector doors and coal flow of the mine hoist.Then the image preprocessing of the collected scene images is studied.Through image preprocessing,the environment noise outside the ROI region is suppressed,and the texture features of the object with recognition are enhanced.The image preprocessing techniques include: region of interest extraction,image graying,image open operation,image filtering and image binarization.By preprocessing the images,the input picture can meet the conditions of the sector doors state algorithm and the coal flow detection algorithm.Then,the two algorithms are further used to identify the images.Secondly,the core module of the sector doors state algorithm is divided into background modeling and template matching.Background modeling can effectively distinguish moving objects,and the doors just meets the detection conditions.There are many algorithms for background modeling.This research uses a mixture of Gaussian background modeling algorithm.Gaussian mixture model can enhance the robustness by moving the center point in multi-modal object recognition.In view of the specific conditions of a certain mine,four state templates are established: the opened storage tank of south and north side and the closed storage tank of two sides.When the door reaches the designated position,the template is matched with the sector doors to identify the state of doors.Coal flow detection algorithm takes foreground frame in advance and stores it.In the process of recognition,the foreground frame is XOR operated with the ROI area of coal flow or coal heap,and the number of non-zero pixels in the obtained image is counted to calculate the proportion of coal flow and coal heap.Finally,this paper combines the two algorithms with the VS compilation platform and the MFC framework,and transmits the detection results and judgment and recognition conditions to the PLC on the mine via Modbus for further research and processing by the mine staff.The thesis has 30 pictures,56 references.