Insulators Extraction In Transmission Line Image Of Patrol Inspection With Helicopter

Transmission lines easily break down because of long distance and regional climate conditions. Insulator, as an important part of a transmission line, is vulnerable to be affected by operating environment.Broken insulator is a serious threat to the normal operation of the transmission line. So far, image segmentation research has been widely researched, but there has been no a universal segmentation method applicable to all of the images. And owing to such many factors as the particularity of the position of the transmission line,limits of imaging equipment,uneven illumination in imaging and so on, the actual background of transmission line image are very complicated, containing many targets such as houses, forests and steel towers etc., which has brought great difficulties in extracting single component. This research mainly adopts image segmentation technique for insulator extraction under complicated background and the segmentation results lays the foundation for the fault diagnosis further.At present, there are few researches on helicopter power-line inspection image segmentation. Through reading a large number of literatures, the combination of Otsu method and genetic algorithm is used for insulator image segmentation considering the characteristics of power-line image. This method overcomes the shortcomings of a simple Otsu method, which has slow calculational time and low efficiency. The optimum efficiency of genetic algorithm will greatly shorten segmentation time. Genetic algorithm is to mainly use maximum between-cluster variance of grayscale image as fitness function.The research will use experimental method to determine the control parameters (population size, crossover probability and mutation probability). In the setting range,multiple tests were carried out on each group data,the optimal parameters are selected by analizing the experiment results. But it is found out in the experiment that the standard genetic algorithm still has some shortages. In order to more satisfy the requirement of real-time of insulator image segmentation, an improved genetic algorithm is used for insulator image segmentation. The improved genetic algorithm is adaptive adjustments crossover probability and mutation probability. The research will isolate the noise by using morphological processing and tilt correction by Hough transform.Finally,the rearch will extract insulator from image and position it in original image.