Research On The Detection And Recognition Of Aerial Insulators Based On RPCA Optimization

Insulator is an indispensable element on transmission lines with the dual function of wire support and electrical insulation.The aerial insulator images are characterized by complex background,low resolution,large in number and more fake target,et cetera.The traditional segmentation method causes user's fatigue and results in bad segmentation quality.At the same time,it is a fault-prone components,its contamination,cracks,damage and other issues will affect the normal use of transmission lines,shorten the line running life,and even lead to tripping power outages,resulting in significant losses.Therefore,it is necessary to carry out real-time detection of the operating state of the insulator.The main work of this paper includes the following three aspects:First,the image pre-processing.due to aerial images in the presence of more noise,complex background and so on,so the insulator image corresponding pre-processing operation.Mainly includes image enhancement and denoising,text information removal.Using image enhancement technology to improve image quality;due to widespread in the aerial image impulse noise and Gaussian noise,so using the filter to remove Gaussian noise.Text information has a certain effect on insulator image segmentation accuracy,and the text information can be removed by Seam Carving method.Through the pretreatment of the image,the contrast and the quality of the image have been greatly improved,the image features are more obvious.Second,the insulator image segmentation,Firstly introduces the problems existing in the aerial insulator image,based on the status quo of insulator special features,is proposed based on the combination of KPCA and shape prior C-V model,a good solution to the insulator and wire adhesion,interference and other issues.And a more accurate segmentation result is obtained.Third,on the basis of the division of insulator,insulator are segmented into multiple umbrella plate image,composition test images.Based on the images of the insulator umbrella images of normal,damaged,cracked and noisy insulators,the sample images of the insulator umbrella are established.The test images were classified by using the convolution neural network.,This method can achieve several different classification of surface fault insulator umbrella images and higher classification accuracy.Insulator effective aerial image segmentation and defect detection as the insulator State detection and fault diagnosis laid the Foundation.