没有合适的资源?快使用搜索试试~ 我知道了~
首页Uniform SAR-SIFT算法文章
资源详情
资源评论
资源推荐
1426 IEEE GEOSCIENCE AND REMOTE SENSING LETTERS, VOL. 12, NO. 7, JULY 2015
A Uniform SIFT-Like Algorithm
for SAR Image Registration
Bangsong Wang, Jixian Zhang, Lijun Lu, Guoman Huang, and Zheng Zhao
Abstract—In this letter, a uniform scale-invariant feature trans-
form (SIFT)-like algorithm is proposed for synthetic aperture
radar (SAR) image registration, which can extract enough robust,
reliable, and uniformly distributed features by the strategies of
optimal f eature selection based on a Voronoi diagram and feature
scale-space proportional extraction. SAR images, taken from dif-
ferent viewpoints by an airborne sensor and at different times by
spaceborne sensors, were used as test data to validate the effec-
tiveness of the proposed algorithm. The indexes of local density
and global coverage were used to assess the spatial distribution
of matches. Compared with the traditional SIFT-like algorithm
for SAR images (SAR-SIFT), the results show that the proposed
algorithm can increase the number of matches and optimize their
spatial distribution.
Index Terms—Image registration, scale-invariant feature trans-
form (SIFT)-like algorithm, synthetic aperture radar (SAR),
Voronoi diagram.
I. INTRODUCTION
I
N synthetic aperture radar (SAR) and interferometric SAR
(InSAR) applications such as 3-D reconstruction, classifi-
cation, and change detection, SAR image registration is still
a challenging task because it is somewhat difficult to register
images in different configurations [1], [2]. The existing regis-
tration approaches can be generally categorized into two major
categories: area-based and feature-based methods [3]. Com-
pared with the area-based methods, the feature-based methods
are recommended in remote sensing, owing to their excellent
performance [4]. Among the feature-based methods, the scale-
invariant feature transform (SIFT) [5] is the classical algorithm
for feature detection and description. It is very suitable for
optical remote sensing image registration [6]. However, differ-
ing from optical images, SAR images are usually corrupted by
speckle noise. The correlativity of adjacent pixels in SAR im-
ages is also much stronger than in optical images. In this case,
the SIFT method does not work for SAR image registration.
Modifications of the SIFT algorithm for SAR images have
been already proposed in the literature. Schwind et al. [7]
Manuscript received November 16, 2014; revised January 14, 2015; accepted
February 13, 2015. Date of publication March 9, 2015; date of current version
March 24, 2015. This work was supported by the Natural Science Foundation
of China under Grant 41401530.
B. Wang is with the School of Resources and Environmental Sci-
ences, Wuhan University, Wuhan 430072, China (e-mail: bangsongwang@
yahoo.com).
J. Zhang, L. Lu, G. Huang, and Z. Zhao are with the Institute of Pho-
togrammetry and Remote Sensing Beijing, Chinese Academy of Surveying and
Mapping, Beijing 100830, China (e-mail: zhangjx@casm.ac.cn; Lulijun2002@
gmail.com; huang.guoman@casm.ac.cn; zhaozheng@casm.ac.cn).
Color versions of one or more of the figures in this paper are available online
at http://ieeexplore.ieee.org.
Digital Object Identifier 10.1109/LGRS.2015.2406336
proposed SIFT-OCT to reduce the influence of speckles by
skipping the first octave of the scale space. While such a pro-
cedure did decrease the number of false detections since many
occurred at that scale, the remaining keypoints were still not
precisely located. You and Fu [2], [3] proposed a bilateral filter
SIFT (BF-SIFT) and an adapted anisotropic Gaussian SIFT
(AAG-SIFT) to find feature matches for SAR image registra-
tion, by replacing the Gaussian scale space with an anisotropic
one. However, the process of both methods decreased image
resolution and caused loss of information. Fan et al. [8] and
Suri et al. [9] suggested to s implify the SIFT by suppressing
the orientation assignment to increase efficiency. However, the
improvement limited the capability of the algorithm to match
images with different viewing conditions.
Dellinger et al. [10] proposed a SIFT-like algorithm specif-
ically dedicated to SAR images, which is called SAR-SIFT,
to register SAR images in different configurations, such as
different incidence angles. SAR-SIFT has performed well on
SAR images, but there are still several problems. The first
problem associated with SAR-SIFT is the lack of controllability
of the number of features, which may produce a redundant
number of features that impose a high computational cost in
the subsequent processing stages, or fewer features than the
required number. Another problem with SAR-SIFT is the lack
of control of the spatial distribution of the extracted features.
It is well known that a uniform distribution of matches is
an essential factor for accurate polynomial transformation, to
deal with the local distortions of remotely sensed images. This
problem might degrade the overall registration quality as a
result of the local geometrical distortions, which is a common
case for SAR images.
In this letter, based on the SAR-SIFT method, we propose
a uniform SIFT-like algorithm for SAR i mage registration,
which is named USAR-SIFT. It can effectively generate enough
robust, reliable, and uniformly distributed features by the strate-
gies of optimal feature selection based on a Voronoi diagram
[11], [12] and feature scale-space proportional extraction [13].
II. SAR-SIFT
Here, we briefly review the entire SAR-SIFT operator pro-
cessing chain to detect and match the keypoints in SAR images.
The SAR-SIFT algorithm consists of three main modules:
keypoint detection, orientation assignment, and descriptor
extraction.
1) Orientation Assignment: Keypoint Detection: The al-
gorithm starts with keypoint detection. For this purpose, a
Gaussian image pyramid is first constructed by convolving
the image with a Gaussian filter. A scale space is then con-
structed by computing the logarithmic ratio of the exponentially
1545-598X © 2015 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.
See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
周梦宝
- 粉丝: 0
- 资源: 3
上传资源 快速赚钱
- 我的内容管理 收起
- 我的资源 快来上传第一个资源
- 我的收益 登录查看自己的收益
- 我的积分 登录查看自己的积分
- 我的C币 登录后查看C币余额
- 我的收藏
- 我的下载
- 下载帮助
会员权益专享
最新资源
- RTL8188FU-Linux-v5.7.4.2-36687.20200602.tar(20765).gz
- c++校园超市商品信息管理系统课程设计说明书(含源代码) (2).pdf
- 建筑供配电系统相关课件.pptx
- 企业管理规章制度及管理模式.doc
- vb打开摄像头.doc
- 云计算-可信计算中认证协议改进方案.pdf
- [详细完整版]单片机编程4.ppt
- c语言常用算法.pdf
- c++经典程序代码大全.pdf
- 单片机数字时钟资料.doc
- 11项目管理前沿1.0.pptx
- 基于ssm的“魅力”繁峙宣传网站的设计与实现论文.doc
- 智慧交通综合解决方案.pptx
- 建筑防潮设计-PowerPointPresentati.pptx
- SPC统计过程控制程序.pptx
- SPC统计方法基础知识.pptx
资源上传下载、课程学习等过程中有任何疑问或建议,欢迎提出宝贵意见哦~我们会及时处理!
点击此处反馈
安全验证
文档复制为VIP权益,开通VIP直接复制
信息提交成功
评论0