基于局部线性转换的无人机影像拼接技术

摘要

According to the big cumulative error,slow speed and other characteristics of unmanned aerial vehicle(UAV)image stitching,in order to improve the accuracy and speed of the splicing,SURF algorithm is used to extract the feature points,a locally linear transforming method is introduced to eliminate false matching points,and at last seamless splicing is completed by feather blending algorithm. The experimental results prove that compared to the classical algorithm,this method increases accurate performance of UAV image matching by about 10％ and improves the speed obviously as well. At the same time,it achieves good performance for image stitching.%针对无人机影像拼接累计误差大,速度慢等特点,为了提高拼接的准确性和速度,本文采用SURF算法提取特征点,引入LLT局部线性转换法来剔除误匹配点,最后通过渐入渐出融合法对图像进行无缝拼接.通过多组实验证明,和经典算法相比,本文方法对无人机影像匹配的准确性能提高约10个百分点,速度也有明显提高,同时可以达到良好的拼接效果.