期刊文献+

单应性约束的SAR图像序列MAP超分辨率重建 被引量:1

MAP super-resolution reconstruction of SAR image sequence with homography constraints
原文传递
导出
摘要 针对合成孔径雷达(SAR)图像序列超分辨率重建过程中对配准误差敏感的问题,该文提出了一种单应性约束的最大后验超分重建方法。首先,对SAR图像序列的中间帧做2倍上采样,将其作为基准图像,利用本文改进的尺度不变特征变换(SIFT)配准算法依次计算SAR图像序列的每一帧与基准图像之间的单应性。通过对待配准图像进行分幅、放大阈值、单应性筛选等操作,达到增加匹配点数量、有效去除误匹配的目的。然后,将单应性作为配准参数,对图像进行配准,并对配准后的图像进行重采样,重采样后的图像利用最大后验(MAP)超分算法进行超分重建,得到高分图像。实验结果表明,该文改进SIFT配准算法可以在保证匹配点对正确率较高的同时增加匹配点数量,且算法复杂度低。改进MAP重建算法与经典超分方法相比,图像质量更高,细节更好。 Aiming at the problem of sensitivity to registration error in super-resolution reconstruction of synthetic aperture radar(SAR)image sequences,a method of maximal posterior super-resolution reconstruction with the homography constraint was proposed.Firstly,the intermediate frames of SAR image sequence were magnified twice,which was regarded as the reference image.The improved scaleinvariant feature transform (SIFT)registration algorithm in this paper was used to calculate the homography between each frame of SAR image sequence and the reference image successively.By means of image segmentation,amplification threshold and homography,the number of feature points could be increased and mismatched could be effectively eliminated.Then,the homography was used as the registration parameter to register each low-resolution image,and the image after registration was resampled.The image after resampling was reconstructed by maximum a posterior(MAP)reconstruction algorithm,and the high-resolution image was obtained.Experimental results showed that the improved SIFT registration algorithm could increase the number of matching points and reduce the complexity while ensuring high accuracy.Compared with the classic reconstruction method,the improved MAP reconstruction algorithm had higher image quality and better details.
作者 卜丽静 苏旭 张正鹏 BULijing;SU Xu;ZHANG Zhengpeng(School of Geomatics,Liaoning Technical University,Fuxin,Liaoning 123000,China)
出处 《测绘科学》 CSCD 北大核心 2019年第8期97-105,125,共10页 Science of Surveying and Mapping
基金 国家自然科学基金项目(41501504) 辽宁省教育厅一般项目(LJYL011)
关键词 SAR图像 尺度不变特征 最大后验 超分重建 SAR image scale-invariant features maximum a posteriori super-resolution reconstruction
  • 相关文献

参考文献10

二级参考文献64

  • 1房建成,全伟,孟小红.基于Delaunay三角剖分的全天自主星图识别算法[J].北京航空航天大学学报,2005,31(3):311-315. 被引量:20
  • 2李晓明,郑链,胡占义.基于SIFT特征的遥感影像自动配准[J].遥感学报,2006,10(6):885-892. 被引量:153
  • 3郭得方.遥感图像的计算机处理和模式识别[M].北京:电子工业出版社,1987.. 被引量:3
  • 4BROWN L G. A survey of image registration techniques [J]. ACM Computing Surveys, 1992, 24(4): 326-376. 被引量:1
  • 5LI H, MANJUNATH B S, MITRA S K. A contour-based approach to multisensor image registration [J]. IEEE Transactions on Image Processing, 1995, 4(3):320-334. 被引量:1
  • 6WANG SH, XIAO J, JIAO L CH, et al.. Fast and accurate automatic SAR image registration using seven invariant moments and improved chain coding of region boundaries [J]. SPIE,2007, 6787:1-7. 被引量:1
  • 7LOWE D G. Distinctive image features from scale-invariant keypoints [J]. International Journal of Computer Vision.2004, 60(2):91-110. 被引量:1
  • 8ZHENG Y, CAO ZH G, YANG X. Multi-spectral remote image registration based on SIFT [J].Electronics Letters.2008, 44(2):107-108. 被引量:1
  • 9EL RUBE I A, SHARKS M A, SALEM A R. Image registration based on multiscale SIFT for remote sensing images. Proceedings of the Third International Conference on Signal Processing and Communication Systems, Omaha, USA: ICSPCS 2009:1-5. 被引量:1
  • 10KE Y, SUKTHANKAR R. PCA-SIFT: A more Distinctive Representation for Local Image Descriptors. Proceedings of International Conference on Pattern Recognition, Washington, USA: ICPR, 2004: 511-517. 被引量:1

共引文献252

同被引文献8

引证文献1

二级引证文献1

相关作者

内容加载中请稍等...

相关机构

内容加载中请稍等...

相关主题

内容加载中请稍等...

浏览历史

内容加载中请稍等...
;
使用帮助 返回顶部