摘要
现有的运动检测算法易受光照、阴影、复杂背景等因素影响,检测结果准确性欠佳,具有一定的局限性。基于此,设计了一种基于Kinect传感器的运动检测方法,可以完成室内场景的刚体运动检测。所设计的运动检测平台基于结合传感器所采集的深度信息和彩色信息,提升物体边界检测精度,提出基于单应性矩阵的刚体运动的密集光流估计方法。实验结果表明,基于Kinect的运动检测平台具有很好的检测效果,证明了运动平台的有效性。
Existing motion detection algorithms are susceptible to illumination,shadows,complex background and other factors,and the accuracy of the detection results is poor,which has certain limitations.Based on this,a Kinect sensor based motion detection method is designed,which can detect the rigid motion of indoor scenes.Based on the depth and color information collected by the sensor,the motion detection platform is designed to improve the detection precision of the object boundary.Finally,the dense optical flow estimation of rigid body motion based on the homography matrix is proposed.The experimental results show that the Kinect-based motion detection platform has a good detection effect,which proves effectiveness of the motion platform.
作者
蔡志兆
吕建勋
CAI Zhizhao;LYU Jianxun(School of Automation Science and Electrical Engineering,Beihang University,Beijing 100191,China)
出处
《应用科技》
CAS
2019年第3期64-69,共6页
Applied Science and Technology
关键词
运动检测
Kinect传感器
运动检测平台
深度信息
彩色信息
边界检测
单应性矩阵
光流
motion detection
Kinect sensor
motion detection platform
depth information
color information
boundarydetection
homography matrix
optical flow
