Research on Image Denoising and Multi-Target Tracking Algorithm in Nuclear Radiation Environment

In this paper, an image denoising and multi-target tracking algorithm suitable for nuclear radiation scenes is designed, and the complexity of the filtering algorithm is reduced through an improved adaptive median filtering algorithm. Reduce the computational load of the network, and adapt to the nuclear radiation scene by introducing the acceleration component into the Kalman filter of the DeepSORT algorithm. Experiments show that the improved adaptive median filtering algorithm can reduce the computing time by about 53% without reducing the filtering effect; the improved YOLOv5s and DeepSORT multi-target tracking algorithms do not significantly sacrifice the tracking effect, and the FPS has been effectively improved. More meet the realtime requirements.