基于扩展卡尔曼滤波的5-Dof圆位姿估计算法  

5-Dof Circular Feature’s Pose Variables Estimation Algorithm Based on Extend Kalman Filtering

作　　者：吴仰玉[1] 李翠 WU Yangyu;LI Cui(Library North Minzu University,Yinchuan Ningxia 750021,China;College of Mechanical Engineering,Beijing Institute of Technology,Beijing 100081,China)

机构地区：[1]北方民族大学图书馆,宁夏银川750021 [2]北京理工大学机电学院,北京100081

出　　处：《传感技术学报》2023年第2期287-293,共7页Chinese Journal of Sensors and Actuators

基　　金：国家自然科学基金项目(11961001);宁夏自然科学基金项目(2018AAC03126);宁夏高等学校一流学科建设(数学学科)(NXYLXK2017B09);北方民族大学重大专项项目(ZDZX201801);宁夏智能信息与大数据处理重点实验室开放基金(2019KLBD004)。

摘　　要：现有圆位姿估计方法对输入帧进行独立处理,忽略了有价值的目标动态信息,圆位姿估计精度有提升空间,提出一种基于EKF的高精度5-Dof圆位姿估计方法,圆位姿由5自由度向量ξ=(X,Y,Z,α,β)^(T)表示。该方法引入贝叶斯框架捕获视频连续帧的时间信息,优化圆位姿估计系统。首先,为了与2D椭圆轮廓交互,算法构造出5自由度向量ξ表示的空间圆投影轮廓5-Dof模型,进而设计非线性测量函数。其次,将该测量函数与扩展卡尔曼滤波(EKF)算法相结合,用于圆位姿优化。此外,使用简单的线性卡尔曼滤波算法(KF)对圆位姿估计值进行修正。实验表明,针对含有不同方差噪声的图像序列,算法利用图像序列的时间相关性,有效提高圆位姿估计精度。A drawback of the existing circular feature’s pose variables estimation algorithms is that input frames are processed independ-ently,so,valuable dynamic information of target is ignored,and the pose estimation accurate can still be improved.A high accuracy 5-Dof circular feature’s pose variables estimation method is proposed by using Extend Kalman filtering,and circular feature’s pose is represented by a 5-DoF(degree of freedom)vector ofξ=(X,Y,Z,α,β)^(T).In this method,bayesian framework is introduced to capture the time information of continuous video frames,and the circular pose estimation system is optimized.Firstly,the proposed algorithm gives projection representation of 3D object circular feature to interact with the ellipse contour in the image by usingξ,and then the non-linear measurement function is obtained.Later,the measurement function and Extend Kalman filtering are combined for pose optimization.In addition,Kalman filtering is utilized to correct pose estimation result,which provides a robust method for object 2D-3D pose estimation by using circular feature in a challenging environment.The experimental results show that the precision of circular feature’s pose estimation is effectively improved by utilizing the time correlated information for image sequence with noises of different square error.

关 键 词：位姿估计 圆特征 空间圆投影轮廓模型 扩展卡尔曼滤波 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]