In view of the problem of misjudgment of switch state image recognition in substation

a kind of method of switch image recognition of substation based on the non-subsampled shearing transform（NSST）image fusion is proposed. The method adopts visible and infrared dual camera mode to obtain visible and infrared image of the switch.According to the characteristics and complementary relationship of them

the visible image and infrared image are fused by using image fusion algorithm based on NSST and the fused image containing rich details and contour features of the two switch images is generated. The improved speeded up robust features（SURF）algorithm is used to extract and match the target characteristic quantity of the fusion image

and then the multi threshold image segmentation technology based on the chaotic cuckoo search（CCS）algorithm is used for processing. Finally

the slope of the line where the switch arm and two contacts are located is obtained on the basis of Hough transformand the switch state is judged in accordance with the angle difference between them. The simulation experiments show that the recognition rate of the proposed method for the disconnector at closing and opening state is 94.4% and 100% respectively and the effectiveness of the proposed method is verified.