Insulator Defect Identification Detection Research Based On UAV Vision

In recent years,power industry of our country has developed rapidly,so the importance of the operation and maintenance of the power system is increasing day by day.One of the important items in the operation and maintenance of power systems is the line inspection.Traditional methods of line inspection are often used to send special staff to check along the route,but this method is inefficient.Therefore,some provinces have attempted to invest in drones to assist the staff in conducting inspections in recent years.The UAVs have good maneuverability and are more secure to the staff.This paper aims to solve the problem of defect recognition of insulator images.During the inspections,UAVs can take pictures of insulators along the way through imagery equipment.Some of these images reflect the defects of insulators.Computers can be used to distinguish the defect insulator’s images by designing an algorithm.This paper proposes a recognition method based on image feature points.Firstly,the original image is converted from the RGB color space to the Lab color space to exclude the effects of light and other factors on the processing result.Then I will segment the insulator part in the image.The morphological processing and area filtering are used to remove the interference of the non-insulator part in the image in the segmentation process.After dividing the insulator part of image,the position of the insulator is determined by means of a straight-line fitting and a model is established based on this to divide the insulator region.The insulator defect condition is determined based on the area ratio of the insulator image in the region.Then an SIFT-based identification method is proposed.Firstly,some background information about SIFT features and their improved PCA-SIFT features are introduced.In the actual test,the SIFT feature extraction effect of the insulator is not ideal,so the extracted PCA-SIFT feature is used and the insulator part in the image is identified using the support vector machine method.These identified insulators are misrecognized,so I use the RANSAC method to optimize the recognition results and eventually to obtain an accurate insulator.Then the insulator’s defects will be tested.The method of detection is to scan the identified insulator and calculate the distance between adjacent insulators.When the distance changes rapidly,it indicates that there is a defect at that location.The two defect detection methods proposed in this paper are compared with the traditional edge extraction defect detection methods.Through the processing of the same batch of images,the actual results of the two detection methods proposed in this paper were tested.Then,usingthe traditional method as a reference,analyze the advantages and disadvantages of the detection method in this paper.According to the results of the comparison,it will guide future research directions.