| Belt conveyors are an important means of transport for the modern mining industry.In mining operations,the hard impurities often mixed in coal,like coal gangue and metal materials,may damage the conveyor belt in the mode of longitudinal tears.Such instantaneous failures normally lead to very expensive downtime of mining equipment and production.Therefore,we need real-time,reliable detection of conveyor belt longitudinal tear.In recent years,visual detection has become an important direction in belt fault detection because it can improve the detection efficiency and precision.Integrated the complementary characteristics of infrared imaging technology and visible light imaging technology,this paper presents a novel method of Integrative Binocular Vision Detection(IBVD)based on infrared and visible light fusion to detect the longitudinal tears of conveyor belts.According to the existing problems of the longitudinal tear visual detection method,the design scheme of the IBVD method is proposed,which inculdes three parts: conveyor belt image acquisition,conveyor image preprocessing,conveyor belt longitudinal tear feature extraction and recognition.Based on the infrared and visible light fusion technology,a new integrated binocular vision sensor device is proposed to collect the infrared and visible light fusion images.In the sensor device,the beam splitter prism is selected as the light splitting element,through a single lens,the incident light is divided into infrared light and visible light,then the infrared and visible light is transmitted into two sensitive element respectively.Infrared signal and visible light signal in the same scene can be captured simultaneously through the same opticalpath at the same time by the proposed sensor device.More important,the rectification is not needed before the image fusion,the fusion image can be obtained immediately.In this paper,the imaging principle of the sensor device is studied,and the simulation of the imaging process is carried out on MATLAB to verify the feasilility and effectiveness of the proposed sensor device.As the sensor device proposed in this paper is still in the theoretical simulation of the stage,so in the conveyor belt image acquisition experimental part,the infrared image and the visible light image are collected by using the infrared camera and the visible light camera respectively.After the registration and fusion of infrared image and visible light image,the conveyor belt infrared and visible light fusion image can be collected.Finally,the image acquisition experiment platform is built in the laboratory to realize the acquisition of the fusion image in the tear state,normal state and scratched state of the conveyor belt.After collecting the fusion image of conveyor belt,in order to make the detection effect more reliable,it is necessary to preprocess the fusion image.In this paper,the characteristics of the conveyor belt longitudinal tear is analyzed.And according to this characteristics of longitudinal tear,a series of pre-processing of the image is carried out,including removing the image noise,enhancing the image contrast,highlighting the area of interest in the tear detection,and dividing the tear target information.After image preprocessing,the image visual effect is better,the tearing target is more prominent.It is well prepared for the feature extraction and recognition of longitudinal tear.On the basis of the pre-processed image,the projection feature of the image is extracted by the projection method.And based on the projection feature,the geometrical features of the image are calculated.Geometrical features are the longitudinal tearing parameters of the conveyor belt,which include the length,width and area of the longitudinal tear.According to the longitudinal tearing parameter and the characteristics of projection feature of conveyor belt,the recognition threshold and the recognition rule of the conveyor belt longitudinal tear is regulated.Finally,the conveyor belt image is classified as tear state,scratch state and normal state,and the recognition of conveyor images with difficult states is realized.The acquired conveyor belt images are processed by the MATLAB programming software,including image preprocessing and tear feature extraction and recognition processing.The experimental results show that the proposed conveyor belt longitudinal tear integrated binocular vision detection method can identify the tear and scratches,and predict the potential tear.The detection accuracy is more than 96% with the detection time less than21 ms,which meet the conveyor belt longitudinal tear real-time online detection requirements.The proposed method is a reliable detection method for the longitudinal tear of the conveyor belt. |
PDF Full Text Request