摘要
无人机自主着舰末端视觉导引中舰机间相对位姿的估测,可以看作机载摄像机对甲板平面3D运动的估测。提出了一种光流分层方法:首先利用已知焦距的机载摄像机拍摄着舰靶标区域的图像序列,并采用Lucas方法计算相邻两帧图像的光流场;而后通过分层模型,将由光流场进行3D运动检测的非线性问题转化为了两个线性问题。该方法无需图像间的特征匹配,可线性解算出着舰靶标区域相对于无人机的三维运动参数,进而得到舰机间的相对位姿信息。计算机合成图和摄像机实拍图像的实验结果验证了该算法的正确性和有效性。
During the last phase of autolanding an UAV （Unmanned Aerial Vehicle） on the deck, the estimation of relative position and orientation between UAV and the naval ship can be regarded as the planar 3D motion estimation between the camera mounted on the UAV and the deck. A hierarchical method based on optical flow was presented. First, the image sequences of landing target were taken by the camera on board with known intrinsic calibration and the Lucas-Kanade method was adopted to compute optical flow of two successive frames. Then a new hierarchical model was proposed to decompose nonlinear problem of the 3D motion estimation into two linear equations. Finally 3-D motion parameters of landing target relative to UAV were recovered without any image feature match. Computer rendered image simulations and experimental results with real video images show that the proposet approaches are both correctness and effectiveness,
出处
《光学技术》
EI
CAS
CSCD
北大核心
2007年第1期102-105,109,共5页
Optical Technique
基金
航空基金资助项目(03I51009)
杰出青年科学基金(50125518)
关键词
无人机
光流
分层方法
视觉导引
平面3D运动估测
UAV
optical flow
hierarchical method
vision guidance
planar 3D motion estimation
