A Design And Implementation Of On-Line Detection System For Longitudinal Rip Of Conveyor Belt Based On Machine Vision

Conveyor belts are widely used for material transportation in all kinds of industries,creating huge economic performance.But its resistance to longitudinal rip is relatively weak,and easily pierced by sharp objects,resulting in long distance crack,causing huge economic losses and even casualties.In this paper,on-line detection system for longitudinal rip of conveyor belt is designed on the basis of in-depth study of existing detection methods.In this paper,feature extraction of the central line of the laser line stripe is converted to feature extraction of the crack with ’-’ shape laser machine irradiating the surface of conveyor belts.When the conveyor belts are no damage,the central line of the laser line stripe is an approximately smooth curve.When the longitudinal rip of conveyor belts occurs,the central line of the laser line stripe are broken.With the detection system tested in actual working condition,the recognition rate of longitudinal rip reaches 90 percent and error recognition rate is reduce to 0 percent.The specific tasks include:Crop the original image and mark the detection area based on prior knowledge to solve the problem of the high volume image data and the low treatment efficiency.According to the gray value distribution differences of the color channels of various areas,an approach for color image gray processing based on linear combination is proposed,so the interference of strong light area and dust area is eliminated in the image.According to relatively small number of pixels in the detection region,an image segmentation method based on average and standard deviation of gray value is presented,so as to segment laser line stripe in the gray image.Morphological operation is used to carry on binary image to acquire laser line mask and strong light mask in the end,so the accuracy of the laser stripe segmentation improves significantly.On the basis of research on a variety of edge detection algorithms,an edge extraction method based on laser line mask is proposed,so that the accuracy of the laser stripe edge extraction improves significantly.After doing research on extraction algorithm of the laser stripe center,the gravity method is used to extract laser stripe center with processing results and time taken into consideration to enhance the extraction stability of the laser stripe center.On account of recognizing rip faults,rip judgment method of the connected line base on concavity and convexity of the curve and rip judgment method based on the altitude difference of the curve are proposed.At last,rip judgment method of multiple image is put forward to reduce error recognition rate of the system from 1.16 percent to 0 percent.