Binocular Visual Pose Measurement of Spatially Invalid Targets with Fused Point-line Features
A binocular vision pose estimation method is proposed,which combines point and line features,and this method is based on the image frame of the video.Firstly,the point and line features of the image frame are extracted through the ORB algorithm and the LSD algorithm,and the LBD descriptor is used to realize Inter-frame matching.Then,based on the epipolar line search method,the features in the left image corresponding to those in the right image are obtained,thereby the spatial coordinates of the matching fea-tures are obtained.Finally,based on the reprojection model of point-line feature fusion,the Levenberg-Mar-quardt method is used to iteratively obtain the location of the failed target's relative pose.The results of simulation experiments show that the visual pose estimation method with point-line features is more robust than the pose estimation method with simple point features,especially for high-speed rotating targets,and it can perform better pose tracking measurement.
Pose measurementBinocular visionSpace invalid targetPoint-line featureEpipolar line search
万方数据
维普
