The working status of porcelain 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 research auto-detection algorithm for insulators. 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 Canny edge detection operator on three aspects: replacement of filter, computation of gradients and thresholds. Afterwards, ways of insulator locating and crack detecting are specified. Based on the new methods above, a crack detection system is designed, put into practice, and proved to be effective and feasible. |