基于单目三维重构的空间非合作目标相对测量

发布时间：2018-04-01 14:03

本文选题：目标重构　切入点：相对位姿测量　出处：《中国空间科学技术》2016年05期

【摘要】：为了解决非合作目标的相对测量问题,提出了一种基于单目图像序列目标重构结果的非合作目标相对位姿测量方法。该方法将目标的三维重构与相机的位姿信息计算相结合,首先利用观测前期得到的图像序列,通过非线性优化算法计算得到目标上部分三维点坐标;然后基于该三维点集合,建立递推深度模型,对相机的相对位姿信息和新观测到的目标点同时进行卡尔曼滤波估计。航拍测量试验表明,随着图片数量的增多,精确重构点的比例(重投影误差小于1个像素的点)不断提高,80%的图像中精确重构点比例优于89%;基于公共数据集的试验表明,该算法对姿态估算精度可达1°以内,位置测量的精度可达到2cm以内。以上试验结果表明,该算法具有较高的测量精度。
[Abstract]:In order to solve the problem of relative measurement of non-cooperative targets, a non-cooperative target relative pose measurement method based on the results of target reconstruction in monocular image sequences is proposed, which combines the 3D reconstruction of the target with the position and attitude information calculation of the camera. Using the image sequence obtained in the early period of observation, the partial 3D coordinates of the target are calculated by nonlinear optimization algorithm, and then the recursive depth model is established based on the set of 3D points. The relative position and attitude information of the camera and the newly observed target point are simultaneously estimated by Kalman filter. The aerial photography test shows that with the increase of the number of images, The ratio of accurate reconstructed points (points with reprojection error less than 1 pixel) has been improved continuously in 80% of the images. Experiments based on common data sets show that the accuracy of the algorithm for attitude estimation can reach less than 1 掳. The precision of position measurement can reach within 2cm. The experimental results show that the algorithm has high accuracy.
【作者单位】：火箭军工程大学控制工程系;
【基金】：国家自然科学基金(61403398) 总装预研基金(9140A01010413JB47005)
【分类号】：V448.2;TP391.41

【相似文献】