摘要
针对现有的基于深度传感器进行指尖检测的工作中不能对任意姿态手掌的指尖进行稳定检测和跟踪的问题,基于卡尔曼滤波器原理并融入对手掌姿态的感应机制,提出一种手掌姿态自适应的卡尔曼单指尖动态跟踪方法.首先借助上一时刻指尖的位置信息得到当前时刻指尖位置的预测值L1;然后通过手掌姿态检测得到用于更新指尖位置的观测值L2;最后根据L1和L2执行更新操作,得到最终的指尖位置估计值L*.实验结果表明,该方法可以实现准确、鲁棒的单指尖位置信息检测和跟踪,能够达到人机交互的实用性标准.
In view of the existing fingertip detection work based on depth sensors, aimed at stable detection and tracking of fingertip of any palm posture, palm posture self-adaption single fingertip tracking is proposed based on Kalman filter theory and by blending in the palm posture sensing mechanism. First get the current predictive value LI of fingertip position according to the fingertip position information of previous moment. Then get the observed value L2 of fingertip position used to update the fingertip position by palm posture detection. Finally get the estimated value L ^* of fingertip position by performing update operation according to L1 and L2 of fingertip position. The results show that the proposed method can realize accurate and robust single fingertip position detection and tracking, and achieve practical standards of human-computer interaction.
出处
《计算机辅助设计与图形学学报》
EI
CSCD
北大核心
2013年第12期1793-1800,共8页
Journal of Computer-Aided Design & Computer Graphics
基金
国家自然科学基金(61070110
61173066)
关键词
普适计算
人机交互
深度传感器
单指尖跟踪
pervasive computing
human-computer interaction
depth sensor
single fingertip tracking