Research On Defect Recognition Of High Voltage Power Cable Joint Based On Image Recognition

The safe and stable operation of power cable lines is the basic guarantee for national production and life.The cable joint is the weakest point of the cable line that is most prone to failure.In order to ensure the reliability of the power supply of the cable line,it is necessary to ensure the installation and construction quality of the cable joints so that there are no minor defects.This is the standard that the construction quality of the cable joints must meet.In this subject,image recognition is added during the construction of cable joints,and image recognition technology is used to detect defects on the cable surface during the joint manufacturing process.If there are defects,the defects are eliminated according to the position of the defect.This paper uses image processing algorithms to detect defects in the cable image.The simulation experiment verifies that the method in this paper can detect minor defects on the main insulation surface,accurately identify the defect category,and the flatness of the peeling and cutting of the external semi-conductive layer can also be better detected.In the image preprocessing part,the non-local mean filter is used to denoise,which has a good denoising effect and maintains good image detail information.In the research of uneven illumination correction,an improved Gamma correction method is proposed.This method can only correct the amount of light.The set correction parameters are related to the overall average brightness of the image,completing the correction of each pixel in the image.The adaptive correction will not affect the detailed information such as the local edge of the image.Simulation experiments verify that this method can perform better correction processing on the image with uneven illumination.In the research of main insulation surface defect detection algorithm,an improved morphological edge detection algorithm is proposed.The algorithm removes noise by opening and closing operations of grayscale morphology,enhances the image,uses multidirectional structural elements to extract defect edges,and adaptively determines the weights of structural elements in each direction.Compared with the edge detection results of traditional differential operators,it has good anti-noise performance and does not have the characteristics of false edges.The algorithm is used to locate the defects on the main insulation surface.The edge details of the positioning by this method are maintained well.The simulation experiment results show that this method can accurately segment all kinds of scratches and impurities,and is suitable for defect detection on the main insulation surface.When detecting defects in the main insulation image,the algorithm in this paper firstly judges whether there are defects in the main insulation image,and if there are defects,then the defects are identified.By extracting geometric shape features from the binarized defect image,the feature is representative of the image,and the trained SVM classifier is used to identify the defect image.The experimental simulation results show that the recognition accuracy is high,indicating that the main insulation of this article Defect detection and recognition algorithms have better performance.This paper designs the software and hardware parts of the cable connector defect detection system.In the design of the image acquisition unit structure,combined with the actual structure of the cable connector,the method of using three cameras to surround the surface of the cable is used to collect the image.The functions realized by the software part of the defect detection system are analyzed,and the algorithm flow of defect detection is elaborated.