一种改进的无人机影像拼接粗差剔除算法

利用无人机技术可以获取高分辨率影像。为了获取高精度的变换矩阵，提高影像匹配效率，本文对RANSAC算法进行了改进，加入影像的灰度信息进行约束，进一步剔除匹配粗差，最后采用均方根误差进行质量评判。为了验证算法的可靠性，选取一组山区影像和一组具有旋转偏角的建筑物影像进行验证。验证结果表明，匹配点粗差剔除率分别提高了15.15%和23.22%，本文算法的均方根误差较小，精度有显著的提高。

An improved algorithm of UAV image mosaic gross error elimination

High-resolution image can be obtained by UAV technology.In order to obtain the high-precision transformation matrix and improve the efficiency of image matching,this paper improves the RANSAC algorithm and adds the gray information of the image to constrain,which further eliminates the coarse error of matching.Finally,it uses the mean square root error to evaluate the quality.In order to verify the reliability of the algorithm,a group of mountain images and a group of building images with rotation angle are selected to verify.The results show that the rejection rate of matching points coarse error is increased by 15.15%and 23.22%,respectively.The mean square root error of the new algorithm is small and the accuracy is significantly improved.