Catenary Insulator Fault Detection Based on Gradient Image Fusion

To accurately identify a single infrared or visible image under all weather conditions, a gradient image fusion model is proposed to fuse infrared and visible images. First, the accelerated up robust features algorithm is used to match the feature points of the two images. Further, the sampled shear wave transform (NSST) algorithm decomposes the image to be fused to form a map with high-frequency and low-frequency component information and then fuses the high-frequency and low-frequency component maps of insulators to achieve local fusion. The high- and low-frequency component maps are inversely transformed by the inverse transform of the NSST to obtain the final fusion map and achieve global fusion. The quality of the fused images is also evaluated. The line fitting algorithm based on the least-squares method is used to detect insulator self-explosions based on a binary image, the pixel integral projection method is used to detect cracks in the insulator, and color features are used to detect whether the insulator surface is polluted. The accuracies of the detection results of a single image and fusion image were compared through experiments. The experimental results show that the recognition rates of insulator self-explosion, insulator cracks, and insulator surface contamination based on fusion images are 95%, 91%, and 90%, respectively, which are higher than the recognition rates of single infrared image or visible image.