在对运动人体进行跟踪时,为保证系统的实时性和精准性,需要根据目标当前的运动轨迹预测目标在下一时刻的位置,并对该时刻的位置进行修正.笔者在卡尔曼滤波理论的基础上,对原有系统的增益矩阵和常数矩阵提出了新的确定方法,简化迭代计算的步骤,提高系统的计算速度,把该理论运用到人体跟踪中,使跟踪速度更为快捷,实现系统的实时跟踪.
When the moving people is tracked, in order to ensure the real time and accurateness of the system, we need predict the new position of the object according to the current moving orbit and amend the position. Based on the Kalman filter algorithm, a new method is proposed to determine gain matrix and constant matrix of the system, so it can simplify the step of the converge calculation, and a real - time tracking system can be realized with the method.