The Research Of Transmission Line Icing Identification System Based On Binocular Vision

With the rapid development of power industry, the voltage level of transmissionline continuously improve, and the distribution has become increasingly wide, so thetransmission line will be confronted with all kinds of challenges, such as the weatherconditions and the geological conditions. Transmission line accident caused by icinghas occurred frequently in recent years, therefore, how to extract the icing informationof transmission line has intrigued widely research. This thesis researched the binocularimage to measure the thickness of icing in transmission line because of the deficiencyof the existing methods.The main content of the algorithm based on image is that the diameter of thepower line increases after icing, the increasing amount of diameter reflects the icingthickness of the power line. In this thesis, we firstly introduce the existing imagemethods which are used to detect the ice-thickness of transmission line, and analyzetheir advantages and disadvantages. A method based on binocular image is adopted todetect the ice thickness of the transmission line.According to the characteristics and requirements of the task, a binocular imageprocessing system is adopted based on DSP. Two high performance CMOS sensors arechosen to take the image of the transmission line. Then the main blocks of the imageprocessing system is introduced.Finally, the algorithm of the image processing that fits the DSP platform isexpatiated in detail. Before image acquisition, an exposure control algorithm is used tomake the image expose appropriately. According to the principles of cameracalibration, a camera calibration algorithm is used to calibrate the binocular cameras.According to the existing matching algorithms and the application environment of theimage system, an algorithm based on edge features guided cost aggregation is adopted.At first, Sobel algorithm is used to enhance the contrast of the image, and LoGalgorithm is used to detect the edge of the image. Then, the edge image is used to guidethe cost aggregation. At last, a disparity map is presented based on WTA algorithm.According to the different distances of the transmission line and its background, thebackground image is easily eliminated in the desired image. In the end, we extract theedge and compare the width between the line covered with ice and the line without iceof the transmission line, then, the thickness of ice is gained. This thesis simulate the icing information of transmission line, and use the detectalgorithm to do the test, and the result is obtained by transplanting the algorithm to theDSP under the Visual DSP++platform. And it can also take some measures to dealwith the irregular ice. It is significant to the research of the transmission line icingidentification.