The Research Of Foreign Body Detection On Transmission Line Based On Image Processing Technology

With the development of economy,the demand for power is increasing year by year.High-voltage lines are the guarantee of electricity supply.Because the transmission line is exposed in wild environment,so that the condition of ageing of line and suspension of foreign body is appeared,which may do harm to the transmission line.It may cause the result of transmission line fracture,which can seriously affecting the normal life of residents and the safe production of enterprises.Traditional way of manual inspection costs a lot of labor,which also restricted by geographical environment and climate change.However,aircraft patrol is an efficient way,which can be used in the long-range and complex environment for implementing the power safety inspection.Aircraft patrol is consist of the control of aircraft and image processing technology.Using airborne camera to collect transmission line video.And automatically analyzes the transmission line status through image processing technology.The automation level of power transmission line is improved by aircraft patrol.At the same time,the accuracy of transmission line inspection is also enhanced.The target detection technology is applied in this paper to analyze the aerial video of transmission lines for solving the problem of trapping foreign bodies,which can achieve the purpose of automatic detection and tracking foreign body on line.Firstly,this paper extracts the key frames of transmission line aerial video.Complex environment factors is considered in this paper,by which the low changing speed of scense information is caused.Gradient lens edge detection algorithm is used to extract key frames.The double threshold shot edge detection algorithm is improved by estimated drift region,by which the purpose of retaining key information in the video can be achieved.Secondly,gray projection method is used to achieve background motion compensation.Least squares predictor is used in this paper to improve gray projection method,which can achieve the purpose of quickly and accurately estimating the amount of background motion.In this way,the amount of deviation between adjacent frames is made up.Thirdly,improvement of three-frame difference method is applied for detecting foreign body in image.Two-dimensional cross-entropy threshold method is used to determine the segmentation threshold in three-frame difference method.By using probability density function,foreign body area can be streamlined.The area of foreign body is marked by eight connected area method.Ultimately,ORB feature points is used to track foreign body in this article.Extracting ORB feature points of foreign body area,which is estimated by Mean-Shift algorithm.Hamming distance is improved by PROMATCH operator,by which the accuracy of feature points can be improved.Estimating the drift region and double threshold method is combined in this paper to extract key frames,by which the amount of calculation can be effectively reduced.And frames contain the representative information.The least squares predictor is applied in the background motion compensation state.Disadvantages of gray-level projection search by frame can be changed.The efficiency of calculation of background is also improved.Applied of 2 d cross entropy in terms of moving target detection can adaptive determine the threshold segmentation in the frame differential method.ORB algorithm is replaced for SIFT and SURF algorithm when tracking the foreign body.Experimental data shows that the algorithm applied in this paper can accurately indicate and track foreign body on transmission lines.The efficiency of indentification of foreign body in also increased.