| Coal is the main energy source in our country. Conveyor belt is the mainmeans of transport and the lifeblood in the coal production. During the beltrunning process, the conveyor belt longitudinal tearing accidents happened oftendue to the steel fracture, external sharp object piercing, mechanical failure,improper operation and other reasons. This leads to casualties and economiclosses. At present, lots of detection methods have been researched aiming at beltlongitudinal tearing phenomenon. The machine detection methods are usedmainly in domestic production, but the detection accuracy is low and thereliability is not high, so these methods cannot meet the demand of real-timedetection. In recent years, visual technology is widely used in industrialinspection field, an online detection system bases on binocular vision for beltlongitudinal tear is designed.The hardware platform including lighting, industrial cameras, optical lens,industrial personal computer, controller and alarm device is built in the system.The system software is designed using C#language based on Visual Studio2010development platform and EmguCV library. The gray images of conveyor beltcaptured by two cameras are processed real-timely by some algorithms such asfiltering enhancement, thresholding, morphological operation and contourfeature extraction. The number, length and area of contour are calculated anddisplayed on the interface, and ultra limit alarming is processed at the same time.The camera parameters setting and frame rate displaying are integrated in thesystem. The experiment results show that the system is real-time and accurate. |
PDF Full Text Request