Airport Pedestrian Target Tracking Algorithm Based on Surf and Improved Correlation Filtering

Tracking target pedestrians in airport terminals has always been a challenge, a pedestrian target tracking algorithm based on surf and improved correlation filtering is proposed. Firstly, the SURF bidirectional matching algorithm is used to automatically identify the surveillance video where the target is located, and the state information of the target in the video image is extracted; Then, the hog feature and the color statistical feature of the target are extracted, and the correlation filter and the foreground background model based on the color statistical feature are trained respectively. The target confidence map is obtained through the feature fusion, so as to predict the location of the target and realize the tracking of the specified pedestrian target. The performance of this algorithm is verified by the comparative experiment with the KCF algorithm; Finally, the effectiveness and real-time performance of this algorithm are verified by the target tracking experiment in the real-time video of the airport terminal.