Target Tracking Method of Transmission Line Insulator Based on Multi Feature Fusion and Adaptive Scale Filter

In view of the complex background of insulator target tracking in transmission line inspection video, which can't adapt to the change of insulator scale, a method of insulator target tracking based on multi feature fusion and adaptive scale filter is proposed. Firstly, a multi feature fusion method combining gray level, color name(CN) and histogram of oriented gradient (HOG) is proposed to extract insulator features to reduce the influence of background changes. Then, the validity of the obtained image is monitored according to the maximum confidence mean ratio, and the size of the filter is adjusted based on the scale factor in the block mode to determine the size of the insulator target. Finally, a classifier is trained by using the minimum decision function to obtain the insulator target position, to improve the accuracy of insulator tracking. The experimental results show that the tracking method proposed in this paper can adapt to the change of insulator scale and environment. Compared with other methods, it has higher tracking accuracy.