POS-assisted and nuclear line-constrained oblique image matching method

doi:10.13474/j.cnki.11-2246.2021.0149

Abstract

Abstract: In order to solve the problem of low efficiency and reliability of UAV oblique image matching based on SIFT feature points due to a large amount of redundant data, large image geometric deformation and repeated texture, a novel image matching method based on POS information and epipolar constraint is proposed. In this method, firstly, the image POS data obtained by the airborne GNSS/IMU device is used to calculate the object overlap area between the images, and then the object overlap area is projected to the image side, and then a highly reliable image pairs are selected according to the image overlap ratio of the two images. Secondly, after using the SIFT-GPU algorithm to extract feature points from the image, the antipolar relationship between the image pairs is estimated based on the POS data. Thirdly, the Euclidean distance between descriptors is a similarity measure, which can achieve efficient and robust matching of feature points. Finally, RANSAC algorithm is used to eliminate mismatches. The reliability of the proposed method is verified by the matching experiment of oblique image.

Key words: oblique image matching, POS data, epipolar constraint, RANSAC algorithm, match pair selection

CLC Number:

WANG Ting, CHEN Chaoran, PENG Mengyuan, HUANG Ershuang, HUANG Xia. POS-assisted and nuclear line-constrained oblique image matching method[J]. Bulletin of Surveying and Mapping, 2021, 0(5): 91-95.

References

[1] 汤国安, 陈正江, 赵牡丹. ArcView 地理信息系统空间分析方法[M]. 北京:科学出版社, 2002.
[2] 沈茗戈. 倾斜三维模型大比例尺城市地形图测绘的应用[J]. 现代信息科技, 2020(13):122-124.
[3] 谢永达. 三维城市模型的构建及其在规划中的应用研究[D]. 武汉:武汉大学, 2004.
[4] 李国柱. 无人机影像构建文物三维模型研究[J]. 苏州科技大学学报(自然科学版), 2017, 34(2):63-66.
[5] 张卡, 盛业华, 付素霞,等. 基于物方定位一致性约束的光学航空影像多视铅垂线轨迹匹配[J].光学精密工程, 2018, 26(7):1784-1792.
[6] 江万寿, 郑顺义, 张祖勋,等. 航空影像特征匹配研究[J]. 武汉大学学报(信息科学版), 2003, 28(5):510-513.
[7] 纪华, 吴元昊, 孙宏海,等. 结合全局信息的SIFT特征匹配算法[J]. 光学精密工程, 2009, 17(2):439-444.
[8] 纪松, 张永生, 范大昭,等.基于特征点引导的多视影像择优匹配方法[J]. 武汉大学学报(信息科学版), 2018, 43(1):37-46.
[9] 邱建国, 张建国, 李凯. 基于Harris与Sift算法的图像匹配方法[J]. 测试技术学报, 2009, 23(3):271-274.
[10] 毛敏. 基于宽基线的弱纹理特征点提取与描述方法研究[D]. 上海:东华大学,2014.
[11] 杨化超, 张书毕, 张秋昭. 基于SIFT的宽基线立体影像最小二乘匹配方法[J]. 测绘学报, 2010, 39(2):187-194.
[12] 纪松, 范大昭, 张永生,等. 多视匹配MVLL算法及其在ADS40线阵影像中的运用[J]. 武汉大学学报(信息科学版), 2009, 34(1):28-31.
[13] LI Z, GRUEN A. Automatic DSM generation from linear array imagery data[J].International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, 2004, 35(B3):128-133.
[14] MOULON P, MONASSE P, PERROT R, et al. OpenMVG:open multiple view geometry[C]//Proceedings of International Workshop on Reproducible Research in Pattern Recognition.
[S.l.]:Springer, Cham, 2017.
[15] 张春森, 张奇源, 郭丙轩,等. 倾斜影像自适应初始物方面元优化匹配算法[J]. 测绘学报, 2020, 49(1):108-116.