Research on State Prediction Based on Multi-Model Fusion

To improve the precision of three-dimensional hand tracking based on the particle filtering, a novel algorithm using Multi-Model fusion to improve the state prediction of particle filtering is put forward. Through analyzing the experimental data of the virtual interactive platform system, combining with the behavior understanding and description of hand, we firstly construct a model for human hand behavior based on cognitive psychology, and then we establishe the state prediction model based on the sigma points during the process of particle filter tracking. We integrate the two models using weighting fusion as the state prediction model during the process of particle filter tracking, which realizes hand three-dimensional tracking based on monocular camera. The results show that the improved algorithm can achieve better prediction precision for hand gestures and improve the tracking precision in the process of hand tracking compared with annealing algorithm. At the same time, since sigma point calculation is simpler, it also saves running time.