对象随型粗糙面近红外单目视觉测量与基准图像自愈

针对复杂应用环境中近红外单目视觉位姿测量系统偏离预置合作目标对象后，无法完成位姿测量的特殊情况，提出一种以合作目标周边随型粗糙面图像为测量对象的单目视觉测量方法，以及图像受损后的动态自愈方法。通过将实时获取的随型粗糙面图像特征与预存基准图像特征进行匹配计算，完成特殊情况下的应急测量。同时，为减少随型粗糙面图像污染或受损后对位姿测量精度的影响，实时计算污染或受损程度并动态自愈基准图像特征。实验结果表明，以随型粗糙面为对象的位姿测量精度稍低于合作目标对象，但能够满足特殊情况下的应急使用需求，提高了测量系统的鲁棒性；当随型粗糙面图像污染或受损达到70%时，采用自愈处理与未做自愈处理相比，方位角测量误差减少72%以上，验证了基准图像自愈方法的有效性。

Near-infrared monocular vision measurement and reference image self-healing of object random rough surface

In response to the special situation where the near-infrared monocular vision pose measurement system in complex application environments deviates from the preset cooperative target object and cannot complete pose measurement， a monocular vision measurement method based on the random rough surface image around the cooperative target is proposed， as well as a dynamic self-healing method after image damage. By matching and calculating the real-time acquired features of the random rough surface image with the pre stored reference image features， emergency measurement in special situations is completed. At the same time， in order to reduce the impact on the pose measurement accuracy after the pollution or damage of the random rough surface image， Real-time computing calculation of the degree of pollution or damage and dynamic self-healing of the reference image features. The experimental results show that the pose measurement accuracy of the random rough surface object is slightly lower than that of the cooperative target object， but it can meet the emergency use needs in special situations and improve the robustness of the measurement system； When the pollution or damage of the random rough surface image reaches 70%， using self-healing processing reduces the azimuth measurement error by more than 72% compared to not doing self-healing processing， verifying the effectiveness of the benchmark image self-healing method.