Design of Embedded Real-time Target Tracking System Based on KNN Algorithm

With the rapid development of science and technology, more and more high-tech products have been applied to all aspects of life. In the field of security and defense, video monitoring plays an important role. In order to solve the problem of real-time and miniaturization of target tracking system, this paper applies a tracking algorithm based on KNN(k-nearest neighbor) background splitter combined with Kalman filter. When tracking a moving target, it is important to determine the moving target in the initial frame. In this paper, the motion detection method in KNN background segmentation is adopted to automatically initialize the moving target. Kalman filter can predict the motion state of the moving target in the next frame according to the information of the moving target in the current frame, and track the target with the predicted value as the input of CamShift algorithm. The implementation of the tracking algorithm on the embedded raspberry pi platform is beneficial to the miniaturization of smart devices. The experimental results show that the algorithm has better tracking ability, faster operation speed and stronger robustness for moving targets in raspberry pi embedded environment, which improves the performance of the algorithm and gets good effects.