546 CHINESE OPTICS LETTERS / Vol. 4, No. 9 / September 10, 2006
Adjacency effect estimation by ground spectra
measurement and satellite optical sensor synchronous
observation data
Jianwen Ma (
)
1,2
and Xue Chen (
ííí
)
1,2,3
1
State Key Laboratory of Remote Sensing Science, Beijing
2
Institute of Remote Sensing Applications, Chinese Academy of Sciences, Beijing 100101
3
School of Geography, Beijing Normal Universit y, Beijing 100875
Received December 29, 2005
This paper addresses the estimation of adjacency effect of CBERS-2 image. The adjacency effect influences
the digital numb er (DN) value of a pixel by adding surrounding scattering signals and path scattering
signals. Based on the theory of radiation transfer model, a procedure is designed to measure the reflectance
from the surface target materials and the materials in a box, which is 1.5 m above the surface to avoid
upwelling reflectance. The results show that the adjacency effect varies from visible, near infrared and
b ecomes steady within short infrared wavelength region; the adjacency effect weakens with the increase
of distance between testing sites. The adjacency effect of CBERS-2 image is corrected and the quality of
the resulting image is impro ved.
OCIS codes: 290.4210, 110.4850, 110.2990, 120.4820.
The adjacency effect is considered as the major correc-
tion procedure after the atmosphere and BRDF correc-
tions. The main difficulty is in the quantitative mea-
surement and estimation of the uncertainty components
of adjacent influence
[1−3]
. This pap er introduces re-
cent work on adjacency effect including ground spec-
trum measurement and adjacency effect identification
on CBERS-2 image.
According to radiation transfer mo del, when only the
direct solar radiation illuminates a surface without at-
mosphere interfering, the ta rget absorbs a fraction of the
incoming photons, the remaining photons are reflected
back to the space. The radiance measured by the satellite
directly depends upon the target properties. In fact, a
part of sunlight reflected by adjacency areas is scattered
to the target. It becomes part of radiance captured by
the sensor. The phenomenon is called adjacent effect,
which blurs the image
[4,5]
. For a non-uniform surface,
adjacency effect is weak to 1 km or lower spatial resolu-
tion and it can be neglected. As a result, dark pixels look
brighter and bright pixels look darker. Especially when
the spatial resolution is higher than 500 m the adjacent
effect correction should be considered.
Adjacency effect models can be summarized into two
groups: 1) developing empirical formulae based on ra-
diation transfer model and 2) using the atmospheric
point spread function (PSF). The radiation transfer
integral-differential equation does not have strict prac-
tical solutions
[6]
, different empirical formulae on adja-
cency effect have been got with different conditional
approximations
[7−10]
. Items in those algorithms are not
straightforward, or they are too complicated due to large
amount of atmospheric parameters. PSF methods in-
cluding image target method
[11,12]
, Monte-Carlo or op-
timization algorithm simulation
[13,14]
, geometrical optics
computation
[15]
, and radiation transfer simulation
[16]
.
Those results rely on specific image information and
have a large random error; or they are computationally
expensive. Moreover, the two classes of models have a
common defect: neither of them is combined with ground
spectra measurement.
Ground spectra measurement methods are designed
based on the principle of radiation transfer model. Fig-
ure 1(a) illustrates that the illumination of the target is
the combination of two kinds of lights which are directly
transmitted sunlight, and the scattering light after in-
teracting with surrounding areas, that is, the analytical
spectral device (ASD) collects light from areas
1
and
2
. The designed platform box ab ove 1.5 m of the target
prevents scattering lights from adjacent areas, ASD=
1
,
see Fig. 1(b). The platform box is about 0.5 × 0.5(m)in
area, 0.15 × 0.15 (m) in height.
On May 4, 2004, we carried out fieldwork in Guant-
ing (N40
◦
17
, E115
◦
41
). Water, grassland and sands
Fig. 1. Principle of the ground adjacency effect measure-
ments.
1671-7694/2006/090546-04 http://www.col.org.cn
资源评论
weixin_38603259
- 粉丝: 5
- 资源: 922
最新资源
- 茶文化推广系统&java&基于spingboot茶文化推广系统设计与实现
- 144200000.xlsx
- spreadjs添加水印
- 中医药命名实体提取数据集
- 基于springboot+vue的音乐播放系统设计与实现（毕业论文+PPT）
- 基于net8的跨平台高性能IOT边缘采集网关，提供底层PLC通讯库，通讯调试软件等
- Redhat切换其他源
- 轻量级物联网开源平台，6分钟快速部署，千万级设备承载、电信级稳定性 物模型-规则引擎-数据通道-组态页面全流程低代码开发
- 基于协同过滤算法的springboot＋vue的煤矿员工健康管理系统（springboot+vue+mysql+说明文档.zip
- GB2312编码,讲解,字母,数字,汉字
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
