基于SIFT算法的无人机航空遥感影像匹配

当无人机低空飞行获取高分辨率遥感影像时,由于不同摄站点拍摄角度不同,使得建筑物等凸出地面的物体在立体像对上成像时产生投影差,导致物体成像几何形状发生畸变并且出现地面高层建筑物之间遮挡现象严重的问题,从而导致匹配困难,成为影响无人机航空遥感影像匹配质量的主要因素.本文采用对旋转、遮挡、缩放、图像局部灰度变化等都具有较强的稳定性的SIFT算法来进行匹配,并取得了很好的效果.

Unmanned Aerial Vehicle image matching based on SIFT algorithm

When the Unmanned Aerial Vehicle(UAV)flies at low altitude to obtain high spatial resolution images,the geometric distortions and occlusions phenomena of the stereoscopic images are serious because of central projection imaging and high buildings’ projection differences which are caused by different cameras’ shooting position.Besides,the buildings which produce projection difference bring the problem of shadow simultaneously,which makes the gray values of the shadowed regions change locally.As a result,it is difficult to match UAV remote sensing images.Therefore,for the low-altitude photography remote sensing platform—UAV,occlusions,distortions and shadow are the major factors on influencing UAV aerial remote sensing image matching quality.This phenomenon is most obvious in the remote sensing images of cities.By adopting SIFT algorithm,which has strong robustness is rotation,occlusion,scaling and shadow,the paper realized accurate image matching for UAV imagery.