沉浸式虚拟现实系统中头部位姿预测方法

Head Pose Prediction Method for Immersive Virtual Reality Systems

为了最小化运动画面延时对沉浸式虚拟现实系统的影响,提出了一种具有运动画面延时抖动补偿功能的卡尔曼滤波算法,用于对用户头部位姿进行预测.首先对运动画面延时进行分析和预测,然后根据头部运动相关性分析结果选择位姿预测方式,并在头部运动建模和位姿预测时分别利用延时预测结果对延时抖动进行补偿.实验结果表明,相比设备原有预测方法,其在一般情况下的姿态预测绝对误差的均值分别减小45.74%、47.25%和40.96%,最大值分别减小11.49%、26.34%和36.79;位置预测绝对误差的均值分别减小35.94%、45.90%和55.81%,最大值分别减小1.05%、25.60%和44.74%.

To minimize the impact of motion-to-photo latency,we propose a Kalman filter algorithm with motion-tophoton latency jitter compensation to predict the user's head pose.First,we analyze and predict the motion-tophoton latency.Second,we select a head pose prediction method according to the correlation analysis results of the head motion.Further,we use the latency prediction result to compensate for the delay jitter in the head motion modeling and pose prediction.Compared to the original prediction method for the equipment,the average of the absolute error of attitude prediction decreases by 45.74%,47.25%,and 40.96%.The maximum decreases by 11.49%,26.34%,and 36.79.Moreover,the average of the absolute error of the position prediction decreases by 35.94%,45.90%,and 55.81%and the maximum decreases by 1.05%,25.60%and 44.74%.