基于VF-SIFT的单目无人机视频快速三维重建技术应用  

The Application of 3D Reconstruction Technology by the Monocular UAV Video Based on VF-SIFT

作　　者：刘睿 李斌兵[2] 张月 

机构地区：[1]武警工程大学研究生管理大队,西安710086 [2]武警工程大学信息工程系,西安710086

出　　处：《武警工程大学学报》2016年第4期29-33,共5页Journal of Engineering University of the Chinese People＇s Armed Police Force

基　　金：国家自然科学基金项目“黄土丘陵区切沟侵蚀过程的三维数值模拟研究”（41171224）

摘　　要：为完成依托于单目无人机的室外地物三维重建，有效利用尺度不变特征变换匹配算法（Scale—invariant Feature Transform，SIFT）提高重建质量，缩短重建时间，运用多图像平均法平滑噪声，然后运用SIFT算法完成对图像的特征点提取，通过高速SIFT算法（VeryFastSIFT，VF-SIFT）为每一个特征点添加4个角度参数，即SIFT扩展描述符，将特征点进行分类匹配，完成了在大地形条件下单个目标的三维重建工作。实验结果表明，该方法减少了三维重建后模型上的孔洞现象，降低了特征点匹配时间，从而在保证重建效果的基础上极大地减少整个三维重建的时间。In order to complete 3D reconstruction based on Monocular UAV aerial, we use SIFT effectively to improve the quality of the reconstructive images and shorten the run time. Aiming at noise, we use the algorithm of image noise filtering. SIFT is applied for ex- tracting feature points. It is assumed that 4 pairwise independent angles can be assigned to each feature and different feature points according to the different mark are matched. 3D re- construction of a single object is completed in an outdoor environment with much lower con- sumption of time. The experimental results show that this method reduces the void ratio and matching time of 3D reconstruction.

关 键 词：计算机视觉 无人机 三维重建 VF-SIFT 

分 类 号：E919[军事]