红外与激光工程
紅外與激光工程
홍외여격광공정
INFRARED AND LASER ENGINEERING
2013年
3期
709-715
,共7页
位姿测量%计算机视觉%在轨服务%非合作目标%多像机非共视场
位姿測量%計算機視覺%在軌服務%非閤作目標%多像機非共視場
위자측량%계산궤시각%재궤복무%비합작목표%다상궤비공시장
pose measurement%computer vision%on-orbit servicing%non-cooperative target%multiple non-overlapping cameras
以空间非合作飞行器视觉位姿测量为背景,针对近距离及超近距离情况下由成像空间小、像机视场等限制,位姿测量所用的视觉特征将不能在单像机中完整成像而无法完成定位的问题,提出一种多像机非共视场的非合作飞行器位姿测量方法.首先将多个像机配置成非共视场的形式,标定各个像机之间的位置关系;然后利用多个像机对目标上的不同特征成像,来自不同像机的底层信息既有冗余又有互补,为位姿测量提供足够的视觉特征和几何特征;最后结合像机之间的位置关系将各个像机中的特征信息进行融合,进而以闭式解法计算目标位姿.实验结果验证了该方法的有效性以及在大目标近距离位姿测量应用中的优越性.
以空間非閤作飛行器視覺位姿測量為揹景,針對近距離及超近距離情況下由成像空間小、像機視場等限製,位姿測量所用的視覺特徵將不能在單像機中完整成像而無法完成定位的問題,提齣一種多像機非共視場的非閤作飛行器位姿測量方法.首先將多箇像機配置成非共視場的形式,標定各箇像機之間的位置關繫;然後利用多箇像機對目標上的不同特徵成像,來自不同像機的底層信息既有冗餘又有互補,為位姿測量提供足夠的視覺特徵和幾何特徵;最後結閤像機之間的位置關繫將各箇像機中的特徵信息進行融閤,進而以閉式解法計算目標位姿.實驗結果驗證瞭該方法的有效性以及在大目標近距離位姿測量應用中的優越性.
이공간비합작비행기시각위자측량위배경,침대근거리급초근거리정황하유성상공간소、상궤시장등한제,위자측량소용적시각특정장불능재단상궤중완정성상이무법완성정위적문제,제출일충다상궤비공시장적비합작비행기위자측량방법.수선장다개상궤배치성비공시장적형식,표정각개상궤지간적위치관계;연후이용다개상궤대목표상적불동특정성상,래자불동상궤적저층신식기유용여우유호보,위위자측량제공족구적시각특정화궤하특정;최후결합상궤지간적위치관계장각개상궤중적특정신식진행융합,진이이폐식해법계산목표위자.실험결과험증료해방법적유효성이급재대목표근거리위자측량응용중적우월성.
Taking pose measurement of non-cooperative space vehicle as research background, a new pose measurement method using multiple cameras with non-overlapping or slightly overlapping views was proposed to solve the problem that monocular vision-based method can not complete the pose estimation in close and ultra-close range, because of the limitation of the imaging space and the field of view, a monocular camera can not observe all the visual features used for pose estimation. Firstly, multiple cameras were configured in non-overlapping form, and the rotations and translations between cameras were calibrated. Then different cameras photographed the different features on the target, the underlying information from different cameras were both redundant and complementary, which provided adequated visual and geometric features for pose measurement. Finally, all the features information from different cameras were fused according to the relationship between cameras to calculate the pose in closed form solution. The experiment results demonstrate its effectiveness and superiority of the method for pose estimation for large target in close and ultra-close range.