Research On Edge Detection Of Transmission Line Image By Patrol Inspection With Helicopter

Transmission line images by patrol inspection with helicopter have complex background and various targets, this causes great difficulties to edge detection and subsequent identification. This paper gives consideration to the application requirements of the actual project, regards the transmission line images as the experimental data, considers the above-mentioned characteristics of the transmission line images by patrol inspection with helicopter, and studies a variety of edge detection technology. Based on this, this paper focuses on the optimal canny edge detection theory, meanwhile, improves the two shortcomings of the traditional canny algorithm when applied. Canny algorithm is superior to other classical edge detection algorithms; it can obtain good edge detection effect in the case of very complex image background, so it is quite suitable for edge detection in images with complex background.The traditional canny edge detection algorithm uses the Gaussian filer to smooth the original image, while the spatial scale parameter of the Gaussian filter need to be set artificially and fixed. The spatial scale parameter set artificially is difficult to meet the needs of both smoothing the noise and protecting the edge details. In response to this limitation, this paper proposes an adaptive Gaussian filer to replace the fixed-scale Gaussian filter; we apply different scale templates to do Gaussian filtering in different regions of the image, it is better than using fixed-scale Gaussian filter.In the traditional canny edge detection algorithm, the selection of the high and low thresholds is not determined by the image edge features, but determined artificially, so it does not have self-adaptability and lacks automation. In response to this limitation, this paper proposes an adaptive threshold selection method; this method constructs a discrete probability model approximate to the Gaussian distribution according to the gradient magnitude histogram, and then calculates the thresholds according to the discrete probability model. This method is simple, selects the appropriate size thresholds, and is a practical adaptive threshold selection method.This paper compares and analyses the traditional canny algorithm and the proposed adaptive canny algorithm. The experiments prove that the proposed methods can improve the edge detection results and enhance the self-adaptability of the canny algorithm. Edge detection of the transmission line images is able to provide bases for subsequent component identification and fault diagnosis; it is significant to improve the intelligence level of the patrol inspection with helicopter.