The Method Research On Power Lines Recognition In Transmission Line Image

During the national 11th five-year plan, the inspection technology of transmission line by helicopter has been widely applied. Using digital image processing and image recognition technology, distinguishing the wire from the transmission line with a complex background images automatically and accurately has a great significance in the safe operation of the electricity transmission grid and other infrastructure.This paper focuses on the identification algorithms, which identify power lines automatically from the transmission line image. Firstly, four common operators, Robert operator, Prewitt operator, Log operator, Canny operator, of edge detection are introduced and applied in power lines detection. It is found that Canny operator is the most suitable for the detection of power lines. Then according to the linear characteristic of the wire edge, three line extraction algorithms are studied, the first one is the Radon transform, the second is the Hough transform, and the third is the method of phase grouping. Those three algorithms are used to extract the edge of wire in the transmission lines image. The advantages and disadvantages of three algorithms are analysized, the extraction results show that a single extraction algorithm is invalid and cann't identify the wire accurately. Therefore, in this paper, phase grouping and Hough transform algorithm are combined to identify the power lines.According to the distribution and geometric features of wires, the phase group method is improved, then a method based on the improved phase group and Hough transform method is proposed. In this method, the improved phase group is used to extract the image edges, meanwhile in order to mark the point coordinates of wires more accurately, the morphological approach is used to process the wire extraction results, finally, Hough transform is used to identify the wires. The results show that the proposed method can identify wires in a complex context automatically with the advantages of strong anti-noise and short running time. It establishes a good foundation for defect detection and components identification.