1175
AIRBORNE MULTIDIMENSION INTEGRATED REMOTE SENSING SYSTEM AMIRS
AND IT’S DATA FUSION
Ma Yanhua 1 ’ 2 , Jing Zhongliang 1 , Wang Jianyu 2 , Shu Rong 2
'• IAST, Shanghai Jiao Tong University, 1954 Huashan Road, Shanghai, 200030 China - mayanhua@mail.sitp.ac.cn
2 . Shanghai Institute of Technical Physics, CAS, Shanghai, 200083, China - shurong@mail.sitp.ac.cn
WG VII/6
KEY WORDS: multidimensional integrated remote sensing system, hyperspectral images, preprocessing, airborne, image fusion
ABSTRACT:
A type of integrated remote sensing system (airborne multidimension integrated remote sensing system —AMIRS) which can
acquire hyperspectral images ,high special numeric stereo images and position-orientation-data at the same time is developed in 2005
to improve the efficiency of remote sensing system .In this paper the data processing flow of the system is listed briefly .An attempt
of the data fusion to get classified map directly from hyperspectral images and high special images is introduced . The method
workflow include pure pixel judgement by grads , pure pixel spectral diffusing according the edge of high spacial image, linear pixel
decompose with already-known pure pixel spectral which simplified the equation ,and un-pure pixel spectral diffusing . In this article
the image fusion method try to base on rough image register methods that means high register precision are not necessary,to do so,
the radiation correlation of two kinds images synchronously acquired are used . Some of experiment result are listed .
1. INTRODUCTION
A type of high - special - resolution - and-high - spectral -
resolution Airborne Integrated Remote Sensing System -
AMIRS ( Xu Weiming, etc,2008) is developed in Shanghai
Institute of Technical Physics in 2005. The system can obtain
hyperspectral images,high spatial resolution images, GPS/IMU
data to calibration the pose of the plane at the same time
(showed in figure 1).
To make better use of the data of this system, a set of data
processing system is developpd. The data fusione introduced in
this article is an attempt of the united-application of
hyperspectral images and the high special images. Only those
related with image fusion is introducfed.
The Parameters of the raw experiments images is showed
below:
- Spatial pi>ftls:
• 610-660nm, 535-585nm, 430-490nm for
multispectral image, 3 bands
• 450-900nm for hyperspectral image, 124 bands,
spectral resolution better than 5nm
- Instantaneous FOV:
• 0.1m GSD(Ground Sample Distance) for
panchromatic and multispectral image
• 0.6m(across track)* 1.2m(along track) GSD for
hyperspectral image;
All the data that used in this article were collected by the
integrated system. Of course, the data that used in image fusion
are post-processed so some Parameters changed, for example,
GSD for hyperspectral image are about lm*lm after pose
correction. This will not be mentioned later.
Figure 1 the structure of AMIRS
2. THE PREPROCESSING
The two kinds of images are both preprocessed before
fusion .The workflow of the preprocessing is showed in figure
2.
2,1 Image correction
The radiation calibration and pose correction are general
preprocessing for both data to any application. Both kinds of the
images are Radiation calibrated and geometry corrected
respectively on their own characteristics.
Both of the hyperspectral data and high special RGB data are
calibrated in the lab with the same integrating sphere. The
calibration of the two instruments not only is used to calibrate
the image, but also guarantees the radiation correlation of two
kinds images in register.
The images used in this article are not completely geometry