| The insulator is a kind of special insulation control made of ceramic or glass,which can play a supporting role in overhead lines.In the past,insulators were mostly used on power poles and gradually applied to some high-voltage transmission lines in remote areas.Glass insulators often cause zero self-explosion under the influence of the external environment.Traditionally,the position detection of the insulator self-detonation often uses manual methods.With the development of UAV technology,the UAV has been used to inspect the insulator.At present,some scholars have used image processing to detect the location of insulator self-detonation,but they have not studied the position of the tower where the self-exploding insulator is found automatically.How to detect the position of the insulator abnormal image and find the position of the tower is a key issue.The UAV cruises according to the navigation path and arrives at the tower position.After the insulators are captured by the fixed point,the video will be stored and transmitted.Combining image processing with UAV is used to meet the requirements of the insulator inspection work.That is,the insulator image is subjected to self-detonation position detection and the position of the tower where it is located is finally determined.The main work and conclusions are as follows:(1)According to the drone flight characteristics and cruise requirements,I design a four-rotor UAV inspection program.According to the characteristics of the insulator image collected by aerial photography,the image contrast is enhanced and image noise were removed separately.Several noise removal algorithms are analyzed and compared with the experimental results to propose an optimal algorithm.That is,the noise removal method based on NSCT-BF.The noise removal effect is the best,the noise can be filtered while retains the edge information of the target.(2)The segmentation algorithm of aerial insulator images was studied.The insulator color features were analyzed,and the segmentation algorithm was used based on the HSI color space conversion.For insulator images with indistinct color features,region-based and edge-based image segmentation algorithms are performed.The optimal segmentation algorithm is improved Canny edge detection algorithm.An adaptive Gaussian filter is instead of the traditional filtering,and the maximum value is suppressed after the gradient calculation.The bimodal histogram is used to improve the threshold and the better segmentation result is obtained.(3)Several identification algorithms were analyzed.The SIFT operator and SURF operator are used to match the segmented insulator images.The affine transformation and SANSAC algorithm are used to eliminate the mismatches.Experiments show that the highest matching recognition rate is based on affine transformation.But matching recognition has a large amount of calculation based on feature points.After the HOG feature extraction of the insulator profile,BP algorithm and PSO-BP algorithm were used for training to improve the recognition rate.(4)An insulator image self-exploding position detection method is introduced and an automatic positioning software for insulator self-detonation position is designed.The identified insulator image is analyzed and modeled to determine the principal axis direction,height and width of the insulator string.A single insulator in the insulator image is extracted as a template.The starting point of the insulator string position in the image is determined,and the template is used to match the histogram of the insulator image.By performing image processing on the aerial insulator image,the insulator self-exploding position detection on the image is completed.The algorithm is transplanted into the software.The experimental results show that the position of the pole tower can be displayed at well as the location of the insulator image self-explosion was detected in the software.This helps the staff to find the damaged tower insulator quickly and replace it. |
PDF Full Text Request