Research On Target Recognition And Tracking Of Insulators Based On Texture Features

The working status of insulators guarantees the safe-running of power system. Because the supervisory and control of insulator state is very important for intelligent control of power system in future, it is significant to development patrol power transmission line system with UAV. In this thesis, several methods to preprocess insulator images are firstly confirmed by analyzing different kinds of image preprocessing techniques, which are graying pictures, boosting up contrast and eliminating noise.Then an effectual edge detection algorithm is newly brought on. This algorithm improves the original edge histogram descriptor of MPEG-7 on two aspects: extraction step and statiscal process. Then a real-time object tracking algorithm based on Kalman filter is proposed. The possible position of the object in the next frame is predicted by Kalman filter's predictive function, and the global searching of an image is converted to local scope, thus the system efficiency and accuracyis improved. Based on the new method above, put into practice, and proved to be effective and feasible.