A video object tracking algorithm combined Kalman filter and adaptive least squares under occlusion

Object occlusion is widespread in video surveillance due to the influence of angle and environment, which brings a great impact on the target tracking and makes the development of video object tracking encountered meeting many constraints. The challenge in video object tracking is how to track accurately when the target is obscured. We use the frame difference method to detect the target and smooth the target trajectory by using Kalman filter. When the target is occluded, we can select the appropriate fitting section after smoothing trajectories and use the least square method to fit the target motion path adaptively, so we can predict the object location. By comparing the experimental results with the actual motion of the target, it shows that the algorithm can be used to track most of the occlusion targets precisely.