Kalman Filter Algorithm for Sports Video Moving Target Tracking

Abstract

In the application of computer vision, the recognition and tracking of sports video targets is an important research topic, which has been widely used in the fields of human-computer interaction, sports video surveillance, intelligent buildings, driver assistants, and image processing. This article introduces feature-based target tracking algorithm, dynamic contour-based target tracking algorithm and filtering theory-based tracking algorithm. Aiming at the problem that the target encountered serious occlusion and the large area of the same color background interference caused the target tracking loss, this paper proposes an improved target tracking algorithm based on adaptive Kalman filtering that can reduce interference. The advantages and disadvantages of the Kalman filter tracking algorithm are analyzed, the multi-feature fusion Kalman filter algorithm is studied, and the weight of each feature is adjusted using an adaptive method. The experimental results show that the improved tracking algorithm has higher tracking accuracy and stable tracking when the target and background colors are similar, the target is interfered by other objects, the target is partially occluded, and the background is more complex.