Algorithm Research And System Realization Of Detection For Longitudinal Rip Of Conveyor Belt Based On Computer Vision

With the continuous improvement of China’s mining capacity, mineral transport capacity has become a key factor affecting productivity. Belt conveyor has widely used in various fields such as mineral development, delivery logistics, cargo handling and so on, because it has lots of advantages, including simple and flexible construction, easy installation; long distance and reliable transport; low power consumption and high efficiency; easy integration, high automation and so on. It has gradually become an ideal conveyor. But once the tearing accident happened because of Stack, friction, overload and other reasons and the conveyor does not stop immediately, the entire belt valued millions will be ripped in a few minutes which will cause unevaluated indirect losses. Longitudinal tearing is one of the most devastating accidents which had happened in several coal mines, so it is very significant to find a reasonable and feasible prevent measures. This article chooses highly targeted image processing algorithm based on machine vision system including image preprocessing, features extracting and recognizing to judge the belt whether has been ripped or not. It can detect the accident at the beginning, so that reduce the losses to lest.In this paper, the main research is done as follow:(1)By knowing the current longitudinal rip detection technique home and abroad, analyzing the main reasons for the longitudinal tearing of belt and combining the advantages of different detection methods, machine vision technique, this paper selects the a method based on computer vision. (2) showing the main structure and principle of the whole detection system for longitudinal rip of conveyor belt. The model of hardware and version of software are selected and designed optimizingly.(3)making deep algorithm research on the image processing and recognizing for the core part of system, combined with the crack features of the conveyor belt longitudinal tear image. Therefore, systematic alarming is achieved relied on two kinds of methods: detection algorithm based on morphologic operator and detection algorithm based on fusion of multi features evidence theory including template matching.(4)The detection algorithms are programmed by using OpenCV image processing library to meet real time feature and improve running efficiency.(5)In the aspect of the whole system software implementation, firstly, the dynamic collection of systematic image realized by API functions which are provide by industrial camera manufacturer; secondly, the OpenCV image processing library and MFC are combined excellently to realize detection algorithm; finally, various of functions are integrated and systematic software is released by using framework based on dialog in MFC.The new theory research and the development of software are combined in this study. The real collection conveyor belt tearing images are used in the procedure of simulation. The dynamic image processing algorithm meets the requirement for on-line detection system and alarming for longitudinal rip of conveyor belt, and protects the real-time of whole system. At the same time, this system also has other advantages, such as high cost-effective, simple structure, strong feasibility, friendly interface and so on, and also provides new thought for longitudinal rip of conveyor belt detection. According to analyzing the experimental results, the detection system is achieved the expected goal.