To obtain a large number of infrared images of insulators, we designed and carried out infrared imaging tests on 35kV line insulators facing environmental interference. Based on these tests, an improved adaptive median filter method is proposed to eliminate the noise in the grayscale. For image enhancement and segmentation, double-threshold Ostu with histogram equalization are combined. A genetic algorithm is introduced to optimize the calculations, and a BPNN recognition model for a zero-value insulator was designed and tested Overall, we conclude that the algorithm and the BPNN model proposed in this paper are effective.

BP neural network,environmental interference,Image Processing,infrared detection,zero-value insulator