中文摘要：

以无人飞行器下视光学成像平台为研究背景,提出了一种相机半标定的下视图像极线校正方法.先对机载下视摄像机的成像几何模型进行了分析,在此基础上推导出一种新的极线校正单应估计函数,并用最小二乘法估计变换参数.结合惯导参数分析可知：所提出的方法不依赖于基础矩阵的精确求解,且在理论上只须1对同名点即可求解.实验结果验证了该方法校正精度高,校正速度快,且图像变形明显小于经典的极线校正算法.

英文摘要：

A method to fully consider the downward-look-imaging feature of unmanned aerial vehicles（UAVs）,was proposed to deal with the problem of epipolar rectification in the semi-calibrated case.The motion vision camera model on UAVs was analyzed at first,and then,a minimization about homograph was set up and solved by least square method.With the inertial navigation parameters,the solution does not need explicit computation of the fundamental matrix but only one parameter.Experiments results show that the method works well both on precision and efficiency,and the image loss is less than state-of-the-art approaches.