Space Dynamic Target Tracking Method Based on Five-frame Difference and Deepsort

Abstract

Abstract For the problem of space dynamic target tracking with occlusion, this paper proposes an online tracking method based on the combination between the five-frame difference and Deepsort (Simple Online and Realtime Tracking with a Deep Association Metric) to achieve the identification first and then tracking of the dynamic target. First of all, with the basis of the three-frame difference, the five-frame difference is designed, and through the integration with the background subtraction of ViBe-based (Visual Background Extraction), the accuracy and anti-interference ability are enhanced; Secondly, the YOLOv5s (You Look Only Once) is improved using preprocessing of DWT (Discrete Wavelet Transformation) and injecting Global Attention Module (GAM), which is considered as the detector for Deepsort, and while solving the loss for target easy to in occlusion situations, the real-time and accuracy can be strengthened; Lastly, in contrast with other methods and cross-validation experiments of datasets, the improved method in this paper is verified for effectiveness and superiority. Simulation results show that the proposed space dynamic target tracking can keep stable to track all dynamic targets under the occlusion, and the tracking precision is improved to 93.88%, accuracy of 71%. Finally, there is a combination with the physical depth camera D435i, experiments on target dynamics and occlusion show the effectiveness and superiority of the proposed recognition and tracking algorithm in the face of dynamic and occlusion.