低空无人机摄影测量影像匹配算法比较研究

Comparative Study of Photogrammetric Image Matching Algorithms of Low Altitude UAV

无人机摄影测量获取的影像具有高分辨率、纹理特征明显等优势,广泛应用于测绘行业各个领域。针对低空无人机获取地物特征不一致的区域,采用SIFT、SURF、ORB、AKAZE 4种影像匹配算法对4类不同地物特征图像进行匹配,深入分析不同算法对不同地物的匹配性能及特点,并通过改变图像旋转角、亮度和增加噪声等方法评价4种算法的稳定性。结果表明:SIFT和SURF在匹配图像过程中稳定性更高。

The images obtained by UAV photogrammetry have advantages of high resolution and obvious texture characteristics,which are widely used in various fields of surveying and mapping industry.In view of the areas where the low altitude UAV acquires inconsistent features of ground objects,four image matching algorithms,namely SIFT,SURF,ORB and AKAZE,are used to match four types of different feature images of ground objects.The matching performance and characteristics of different algorithms for different ground objects are analyzed in depth.The stability of the four algorithms is evaluated by changing the image rotation angle,brightness and adding noise.The results show that SIFT and SURF are more stable in the process of matching images.