Research On Insulator Defect Detection Technology Based On Infrared And Visible Light Image Fusion

Insulators,As one of the most important equipment in the contact net,the insulator plays the role of supporting the feed line and blocking the current flow to the tower pole.However,the catenary insulator is exposed to the external environment for a long time,and is vulnerable to the influence of rain,snow,haze,sunlight,strong electric field and strong mechanical tension,making the insulator damaged,self-explosion,pollution and other defects.These defects seriously threaten the safe operation of the locomotive.Therefore,it is necessary to detect some common defects of insulators,find the hidden dangers in advance,and eliminate the traffic safety accidents caused by the defect of insulators.Traditional insulator defect detection methods are mainly based on image processing techniques,such as morphology,edge detection,texture analysis,etc.,but these methods have problems such as low detection accuracy and easy interference.Compared with the traditional method,the method in this paper has higher accuracy and robustness in image fusion and defect detection,and can effectively detect self-detonation,crack,filth and other defects of insulators.This study is of great significance to improve the safety and reliability of OCS insulators and guarantee the safe operation of railway traffic.In order to solve the chronic exposure of contact net insulators,a method based on the fusion of infrared and visible images is proposed to perform insulator defect detection.The main research results and innovations are as follows:(1)For the problem that a single infrared image or visible light image can not be detected all-weather,a gradient image fusion model algorithm is proposed to integrate the visible light image and infrared image of insulators together.First,the infrared and visible images are preprocessed,and then the accelerated robust feature algorithm(Speeded-Up Robust Features,SURF)is used to match the feature points of the two images,so that the two images can be aligned in different dimensions,and then determine the mapping relationship between the two,laying a foundation for the fusion of the images.In terms of image fusion,the sampling shear wave transformation(Non-subsampled Shearlet Transform,NSST)algorithm image into high frequency subband and low frequency subband diagram,respectively for high frequency and low frequency component diagram fusion,realize local fusion,reuse NSST reverse transformation of high frequency subband diagram and low frequency subband inverse transformation,get the final fusion figure,realize the global fusion.Therefore,the advantages of infrared image make up for the disadvantages of visible light and increase the ability of the system to resist external interference;(2)in view of the existing neural network algorithm for image fusion is vulnerable to the interference of the external environment and lead to a lot of detail information missing and multiscale transformation algorithm redundant information and operation time to increase problems,put forward the method of combining multiscale transformation and neural network algorithm to fusion processing,complementary advantages between the two.Using a pulse-coupled neural network(Coupled Neural Net-works,The PCNN)algorithm merges the high-frequency subband map of the two images together,Using a weighted local energy(Weighted Likelihood Estimation,WLE)and an integrated optimization algorithm(WLE-WSEML)combine the low-frequency subband map of the two images together,Using multi-scale inverse transformation to transform the fused high-frequency subband map into an whole graph,So as to achieve the global integration;(3)To evaluate the question of whether the use of the image fusion algorithm is reasonable,This paper evaluates the fused image quality by calculating the edge information retention degree,information entropy,mutual information,spatial frequency,mean square error,and structural similarity.(4)For the common problems of explosions,cracks and pollution,three detection methods are put forward: the linear fitting algorithm on the basis of the binary image;and the pixel integral projection method is used to detect the pollution on the surface of the insulator.After experimental verification,the proposed method based on infrared and visible images can effectively improve the defect detection ability of catenary insulators.Compared with the traditional method,the method of fusion image can improve the recognition rate of insulator defect,especially in the detection of insulator self-explosion,insulator sheet crack and insulator surface pollution,the recognition rate reaches 95%,91%,91% and 90% respectively,which are higher than the recognition rate of a single infrared image or visible light image.This shows that the proposed method has higher accuracy and robustness,which can effectively detect the failure of insulators and provides an important guarantee for the safe operation of catenary insulators.Therefore,this study has important practical application value and is important for improving the safety and reliability of catenary insulators.