An Algorithm for Analyzing Group Motion Trajectory Based on Feature Recognition
Aimed at the problems of long execution time and low prediction accuracy in group motion trajectory analysis algo-rithms,this paper proposes a group motion trajectory analysis algorithm with higher prediction accuracy based on Gaussian mixture model,mean shift algorithm and LSTM model.By reviewing and analyzing mainstream group target tracking algo-rithms,a mean shift algorithm based on particle filtering is designed and constructed.Based on this,an LSTM model and a Gaussian mixture model with spatiotemporal prediction function are introduced to achieve anti-interference recognition,tracking and prediction of group target motion trajectories in video environments.The simulation experiment uses the same basketball video dataset as an sample,and the statistical results show that compared with traditional group recognition and tracking algo-rithms,the group motion trajectory analysis algorithm based on feature recognition has higher prediction accuracy and less av-erage execution time.The prediction accuracy of the group motion trajectory is improved by 14.9 percentage points,and the average execution time of the algorithm is reduced by 21.6 percentage points.
万方数据
维普
