摘要
针对手方向估计时的实时性差以及准确率低问题,提出一种基于Kinect的手方向实时估计系统。利用Holt双指数平滑滤波算法来平滑Kinect的骨骼数据,并根据骨骼数据定位到手腕点。通过设置合理的阈值检测俯仰角的状态来解决四元数转换为欧拉角过程中奇异解的问题。提出一种将卡尔曼滤波和中值滤波相结合的算法对欧拉角进行滤波,使得欧拉角更平稳。通过在彩色图像对手方向进行估计以及运用OpenGL构建一个3D手模型来实时地模仿手的运动,验证了该系统具有较强的实时性、准确性和可靠性。
Aiming at the problem of poor real-time performance and low accuracy when estimating the orientation of the hand,the paper proposed a real-time estimation system of hand orientation based on Kinect.We used Holt double exponential smoothing filter algorithm to smooth Kinect’s skeleton data,and located to the wrist point according to the skeleton data.We solved the problem of singular solutions in the process of converting quaternions to Euler angles by setting a reasonable threshold to detect the state of the pitch angle.We proposed an algorithm that combined Kalman filtering and median filtering to filter Euler angles,making Euler angles more stable.By estimating the orientation of the hand in the color image and constructing a 3D hand model using OpenGL to simulate the movement of the hand in real time,it is verified that the system has strong real-time,accuracy and reliability.
作者
姚玲
陈建新
潘招来
黄湘君
Yao Ling;Chen Jianxin;Pan Zhaolai;Huang Xiangjun(College of Telecommunications and Information Engineering,Nanjing University of Posts and Telecommunications,Nanjing 210000,Jiangsu,China)
出处
《计算机应用与软件》
北大核心
2023年第2期186-191,共6页
Computer Applications and Software
基金
国家自然科学基金项目(61901227)
南京邮电大学项目(NY217021,NY218014)。
