摘要
传统图像匹配算法存在精准度低、效率不高等问题。文章在传统图像匹配算法的基础上提出一种新的ORB算法方式,将非线性扩散滤波的标度空间进行算法的重新构建。该方法比高斯标度空间更能有效地保持图像的细节,可提高匹配算法的稳健性。在此基础上,文章利用Fast算法获取特征点,并以特征点为中心选取适当的窗口进行特征点提取;然后采用栅格移动统计方法对内部节点进行选择。ORB试验结果显示,与传统ORB相比,文章研究算法的匹配精度提高了5%,匹配效率提高了65%,说明文章研究算法在航空影像的匹配精度和效率上都有很大的提升,可以很好地用于航空影像的匹配。
Traditional image matching algorithms have some problems,such as low accuracy and low efficiency.In this study,a new ORB algorithm is proposed based on the traditional image matching algorithm,and the scale space of nonlinear diffusion filtering is reconstructed.This method can keep the details of the image more effectively than Gaussian scale space and improve the robustness of the matching algorithm.On this basis,the Fast algorithm is used to obtain feature points,and an appropriate window is selected around the feature points to extract feature points.Finally,the grid movement statistics method is used to select the internal nodes.The results of ORB experiments show that the matching accuracy and efficiency of the algorithm proposed in this study are improved by 5%and 65%compared with the traditional ORB,which shows that the matching accuracy and efficiency of aerial images have been greatly improved,and it can be well used for aerial image matching.
作者
罗伟
Luo Wei(Shandong Foreign Trade Vocational College,Qingdao 266100,China)
出处
《无线互联科技》
2023年第19期121-124,共4页
Wireless Internet Technology
关键词
局部特征
图像匹配算法
ORB算法
栅格移动统计方法
local features
image matching algorithm
ORB algorithm
grid moving statistical method
