首页 | 本学科首页   官方微博 | 高级检索  
     

基于半投影翘曲的无人机图像拼接方法
引用本文:陈珺,高海宽,李梓贤.基于半投影翘曲的无人机图像拼接方法[J].计算机应用研究,2024,41(1).
作者姓名:陈珺  高海宽  李梓贤
作者单位:中国地质大学(武汉),中国地质大学(武汉),中国地质大学(武汉)
基金项目:国家自然科学基金面上项目(62073304)
摘    要:为了解决无人机图像在拼接过程中出现的错位、重影等造成图像失真的问题,提出一种基于半投影翘曲的无人机图像拼接方法。首先估计图像的全局投影变换矩阵;然后根据待拼接图像的重叠情况,以及投影变换和相似性变换的连续性,求解出过渡变换矩阵和相似性变换矩阵,得到最终的半投影变换矩阵;最后构建图像重叠区域的差异矩阵,以此为基础获取重叠区域的差异性区域。使用分区融合策略,在差异性区域进行单采样,其他区域进行距离加权融合。实验结果表明,该方法可以很好地拼接由于视角变化、地势起伏等造成视差的无人机图像,得到的拼接结果自然清晰,效果优于其他先进算法。该方法有效地解决了无人机图像拼接错位、重影的问题,在多项定量指标上表现良好。

关 键 词:图像拼接    无人机图像    图像失真    半投影翘曲
收稿时间:2023/5/7 0:00:00
修稿时间:2023/7/20 0:00:00

UAV image stitching method based on half-projection warping
Chen Jun,Gao Haikuan and Li Zixian.UAV image stitching method based on half-projection warping[J].Application Research of Computers,2024,41(1).
Authors:Chen Jun  Gao Haikuan and Li Zixian
Affiliation:China University of Geosciences,,
Abstract:To solve the problems of distortion and ghosting in the stitching of UAV images, this paper proposed a UAV image stitching method based on half-projection warping. Firstly, it estimated the global projection transformation matrix of the image. Then it solved the transition transformation matrix and the similarity transformation matrix according to the overlap of the images, the continuity of the projection transformation and the similarity transformation, and obtained the final semi-projection transformation matrix. Finally, it constructed the difference matrix of the overlapping region of the images and obtained the different regions of the overlapping region based on the difference matrix. Using the partition fusion strategy, it performed single sampling in the different regions and distance-weighted fusion in other regions. The experimental results show that this method can splice the images caused by parallax due to the change of perspective and relief, and the result of the splicing is natural and clear, and the effect is better than other advanced algorithms. The method effectively solves the problems of misalignment and double image stitching of UAV images and performs well in many quantitative indexes.
Keywords:image stitching  UAV image  image distortion  half-projection warping
点击此处可从《计算机应用研究》浏览原始摘要信息
点击此处可从《计算机应用研究》下载全文
设为首页 | 免责声明 | 关于勤云 | 加入收藏

Copyright©北京勤云科技发展有限公司  京ICP备09084417号