中文摘要：

为提高无人机（UAV）自主着陆的能力，提出了一种新方法，仅依靠视觉传感器数据，基于空时处理估计UAV的俯仰角和高度参数，构建双目立体视觉导航系统采集视频数据，利用Hough变换和RANSAC方法从单目图像序列中计算灭点；再利用灭点几何模型计算无人机着陆的俯仰角，基于双目图像序列，提取Harris角点进行特征点匹配，获取无人机的深度信息，联合无人机的俯仰角与深度信息，基于三维重建方法获得无人机的高度参数，根据UAV的运动规律，采用自适应卡尔曼滤波进一步提高UAV高度估计的精度，模拟实验结果表明，所提出的方法可以有效地估计UAV着陆的俯仰角和相对跑道的高度参数，而且算法具有较快的收敛速度。

英文摘要：

A novel approach was presented to improve the autonomous landing capability for unmanned aerial vehicle （UAV）. The approach estimated the pitch attitude and altitude of UAV sole dependence on the data from visual sensors by spatial-temporal processing. A stereo visual navigation system was built to record the video data. The vanishing point from monocular sequences was extracted by Hough transform and RANSAC algorithm, and a geometric model of vanishing point was presented to estimate the pitch attitude of UAV. The feature-based matching method was adopted to obtain depth information by extracting Harris corner from stereo sequences. The altitude of UAV was achieved by three-dimensional reconstruction with the combination of pitch attitude and depth information. An adaptive Kalman filter algorithm was proposed to improve the precision of altitude estimation based on the UAV motion characteristic. Simulation results show that the approach can effectively estimate the pitch attitude and the height relative to the runway, and converge quickly.