An Improved Particle Filter Tracking Algorithm with Background Information Fusion

The traditional particle filter tracking algorithm usually leads to tracking error or failure, when the target is interfered by the similar background or blocked by the other object. To eliminate these effects, an improved method was proposed. To overcome the background interference, the logarithm likelihood function was used to fuse the background information into the target model, and the target was divided into multiple sub-regions to fuse the spatial information into the target model, which increase the model reliability. Considering that the target might be under occlusion, the historical trajectory information was stored over a little of time and the least square method was used to predict the target location in the next frame. Experimental results show that the improved tracking method is more robust to background interference and has better tracking accuracy in occlusion cases than the traditional particle filter.