通过对状态预报误差的协方差矩阵的各个分量分别进行控制,设计一种强跟踪扩展卡尔曼跟踪滤波器,并在单目相机环境下利用该滤波器对3D人手运动过程中的突变状态进行跟踪;讨论了3D人手生理模型、运动模型等问题;最后通过大量实验证明了该强跟踪滤波器的有效性.与现有的STEKF滤波器相比,文中的强跟踪滤波器具有更强的跟踪能力.
By controlling independently each components of the error covariance matrix, a new strong tracking extended Kalman filter(STEKF) is put forward which can effectively track the abruptly changing states of a moving human hand with a single camera. Besides, the design of 3D physiological models and kinematic models of the tracked hand are also discussed. At last, several experiments validate the proposed filter. Compared with the STEKF proposed in [6], our algorithm manifests superiority of tracking abrupt changing states.