Laser-based on-line machine vision detection for longitudinal rip of conveyor belt

Conveyor belts are widely used in mines, power plants, ports to transfer lump materials or products. They are prone to longitudinal rip in operation, which generally results in a stoppage of production, even causes belt break or longitudinal rip accidents for lacking the detection and treatment in time. It has long been recognized that a conveyor belt monitoring system for early detection of longitudinal rip is desirable. Based on the line laser detection technology, an on-line detection method is investigated to accomplish conveyor belt longitudinal rip detection rapidly and accurately, and the monitoring system utilizing machine vision technology is designed in this paper. A red line laser stripe projects on the surface of conveyor belt, and the image obtained by the corresponding CMOS camera is processed and analyzed to judge whether there is a rip on the belt surface or not. Firstly, the red stripe region of interest is segmented. Then, the skeleton representation of the stripe center and the binary image are obtained by maximum pixel value method. And the abnormal pixels, which correspond to the rip position, are detected by the neighborhood search method using the difference algorithm and the curvature method. Finally, the fault region is marked. Experimental results show that the proposed method is fast and high precise. This on-line detection method and system can effectively resolve the problem of conveyor belt longitudinal rip detection, and can be used during production hours with a full conveyor load.