摘要
针对单目视觉定位方法定位精度高但数据源不稳定,而惯性测量组件可稳定获取定位数据却存在累计误差的问题,提出了一种融合惯导信息的单目视觉室内定位方法.该方法利用四参数拟合模型将图像数据转换成定位定姿数据,并在惯导数据解算过程中引入互补滤波修正陀螺仪读数,最后将处理后的数据作为观测值输入到扩展卡尔曼滤波器中,得到最优位置信息.实验结果表明,该方法能够有效地提高室内定位的精度和稳定性.
Aiming at the problem of high positioning accuracy but unstable data sources in monocular vision positioning method,and accumulative errors in inertial measurement unit(IMU)which can obtain positioning data steadily,a monocular vision indoor positioning method is proposed,which integrates inertial navigation information.The method uses four-parameter fitting model to convert the image data into positioning and attitude data,and introduces complementary filtering to correct the gyroscope reading in the inertial data solving process.Finally,the processed data is input to extended Kalman filter.as observation value to obtain the optimal position in formation.The experimental results show that the method can effectively improve the accuracy and stability of indoor positioning.
作者
李梦珍
余敏
俞佳豪
LI Mengzhen;YU Min;YU Jiahao(College of Computer Information and Engineering,Jiangxi Normal University,Nanchang 330022,China)
出处
《全球定位系统》
CSCD
2019年第3期74-80,共7页
Gnss World of China
基金
国家重点研发计划(2016YFB0502204)
关键词
单目视觉定位
惯导
扩展卡尔曼滤波
互补滤波
数据融合
monocular visual localization
IMU
extended kalman filtering
complementary filtering
data fusion
