基于SIFT的航空侦查图像拼接算法  

Image mosaic algorithm of aerial investigation based on SIFT

作　　者：孙艳丽 周伟 张亚周 柳叶 Sun Yanli;Zhou Wei;Zhang Yazhou;Liu Ye(Naval Aviation University,Yantai 264000,China)

机构地区：[1]海军航空大学,烟台264000

出　　处：《电子测量技术》2020年第19期112-116,共5页Electronic Measurement Technology

基　　金：国家自然科学基金(61871392)项目资助。

摘　　要：针对具有一定重叠区域的航空侦察图像进行拼接,研究图像配准及融合。图像配准效果直接影响图像拼接的结果,其核心是特征提取与匹配。采用SIFT算法进行特征提取,通过NN法配对,并采用RANSAC法剔除错误匹配点,实现图像配准,并对重合区域采用渐入渐出融合法进行融合处理,实现了航空侦察图像的拼接。由实验结果可以看出该算法得到的融合图像具有更好的视觉质量,满足应用要求。Focuses on image registration and fusion of aerial reconnaissance images with certain overlapping areas.The image registration effect directly affects the result of image mosaic,the core of which is feature extraction and matching.SIFT algorithm is adopted for feature extraction,NN method is adopted for pairing,and RANSAC method is adopted to eliminate the wrong matching points,so as to achieve image registration,and the overlapping area is fused with fade in and fade out fusion method,so as to realize the aerial reconnaissance image mosaic.The experimental results show that the fused images obtained by this algorithm have better visual quality and meet the application requirements.

关 键 词：航空侦察 图像拼接 图像配准 

分 类 号：TP751[自动化与计算机技术—检测技术与自动化装置]