Block Detection and Tracking Algorithm of Foreign Objects Debris in Airport Runway Based on Background Alignment and Difference

Aiming at the problems of unmanned detection and recognition of foreign object debris(FOD) in the autonomous detection and recognition of airport runways, the detection results are affected by the environment, the detection of small targets is difficult, and the rate of missed detection is high. An airport runway foreign object debris sub-block detection and tracking method based on background difference is proposed in this article. First, the linear interpolation of position information assisted by the speed is used to obtain the most relevant frame of the image in background image template library and the image sequence to be identified. Then, the image is divided into different sub-blocks, the adaptive weights are set using the image texture complexity in each sub-block, and the feature point extraction is combined with the oriented fast and rotated brief(ORB) method to normalize the sub-block feature points to the original image. Align the image with the most relevant frame of the background template library using the feature points result before to obtain the foreign object detection result. Finally, the kernelized correlation filter is introduced to track the detection results for multiple targets. Local tracking methods of each sub-block are used to reduce the computation time and the tracking results are tested for reliability. According to the experiment results in three different scenarios, compared with another five common image-based airport FOD detection algorithms, this method can reduce the overall false detection rate of foreign objects by more than 70%, and the overall missed detection rate to 0 when the size of foreign body is larger than 1 cm×1 cm, while ensuring the algorithm processing speed, which achieved good results.