基于小波变换和Kalman滤波的虚拟人空间位置处理

Processing Method of Space Locations of Virtual Human Based on Wavelet Transform and Kalman Filter

为了实现对虚拟人运动过程的有效控制,针对光学运动捕捉设备捕获的虚拟人空间位置数据平滑性不足和加速度误差大的问题,在剔除了运动数据中错误点的基础上,提出了采用基于小波变换和Kalman滤波的空间位置信息处理方法,通过采用基于小波变换的滤波处理方法获取主要的加速度信息,再基于弱跟踪Kalman滤波技术实现对空间位置坐标和加速度信息的修正和平滑处理。与Kalman滤波方法相比,该方法可获得更为准确和平滑的虚拟人体空间位置和加速度位置信息,为进一步分析虚拟人的运动过程、判断虚拟人操作动作提供了更准确的信息保障。

For the problem of inadequate smoothness of space location coordinate and large error of acceleration information, a processing method based on wavelet transform and Kalman filter was proposed to realize the valid control of virtual human＇ motion process after eliminating error points of motion data. Main acceleration information could be achieved based on the filter method of wavelet transform, and weak tracking Kalman filter was introduced to realize the amendment and smoothness of space location coordinates and acceleration information. Compared to the conventional Kalman filter method, the proposed method could achieve more exact and smooth space location and acceleration information of virtual human and provide a more accurate information guarantee for further analysis of virtual human＇ motion process and estimation of virtual human＇ operating actions.