Extraction And Recognition Of Aerial Video Insulatof Parts

With the gradual deepening of China’s reform and opening up, the rapid development of economic construction, the demand for electricity is also growing with the development of various industries increments. Insulator is an important component of transmission line, so its regular inspection is necessary to ensure the normal operation of the power industry. In the complex regional environment, such as mountains, plains terrain varies, where traditional manual inspection method is not only time-consuming, and with a certain degree of risk. Currently, the helicopter has been widely used in transmission line inspection, how to use the computer vision techniques to process the aerial video or image has become the focus of the study.This paper studies the aerial video on high voltage transmission line, insulator identification and location tracking in order for the latter to provide identification of insulator fault conditions. First, the aerial insulators video preprocessing, the video frame image noise removing, has a mean filtering, median filtering and Gaussian filtering in three ways on the video frame image noise removing, using the peak signal to noise ratio (PSNR) compared noise removing effect, select the appropriate filter.Then, discussed several kinds of target detection method for this model and several kinds of video object, introduces the method of the target model and the target detection in video commonly used:optical flow method, frame difference method and background difference method, and through the analysis and comparison of the experimental results obtained:optical flow method for velocity change objects will be detected, is only suitable for moving objects more obvious video; inter-frame difference method for foreground and background while changing scene is very difficult to distinguish, interference factors can not be removed; the background difference method has a very high demand, the background modeling algorithm complexity, poor real-time, these methods are not suitable for containing abrupt shot and aerial video with complex background. Taking into account the insulator in aerial video color than the tower and background color differences, this paper analysis as a starting point to the target model based on color, the Mean-shift and Camshift object tracking method is introduced and analyzed, through the experimental comparison of extraction and identification of the Camshift target tracking algorithm suitable for insulator of the insulators in video based on the characteristics of the target model, the color is in line with the video. On this basis, through the analysis of video and understanding of Camshift, analyses the improvement, the algorithm has better effect on the extraction and recognition of the insulator.