Ono Tetsu 
  
672 
DIGITAL MAPPING USING HIGH RESOLUTION SATELLITE IMAGERY BASE 
ON 2D AFFINE PROJECTION MODEL 
Tetsu ONO*, Susumu HATTORI**, Hiroyuki HASEGAWA***, Shin-ichi AKAMATSU* 
*Graduate School of Engineering, Kyoto University, Yoshida-Honmachi, Sakyo-ku, Kyoto 606-8501, JAP; 
ono@jf.gee.kyoto-u.ac.jp, akamatsu@Qinfo.gee.kyoto-u.ac.jp 
**Faculty of Engineering, Fukuyama University, 1 Sanzo, Gakuen-cho, Fukuyama, 729-0292, JAPAN 
hattori@huip.fukuyama-u.ac.jp 
***Geonet Inc, 1265-1-203, Shin-yoshida, Minato-ku, Yokohama, 223-0056, JAPAN 
hasegawa@geonetz.com 
KEY WORDS: High-resolution satellite imagery, Topographic mapping, Digital ortho-imagery, 2D aff 
projection model 
ABSTRACT 
High-resolution satellite imagery is expected to reduce cost for medium and small scale topographic mapping 
Because 1m high-resolution satellite imagery has a much narrower field angle, the projection of images is neal 
approximated by parallel rather than central one. In this situation, the orientation model based on affi 
projection is effective for satellite imagery triangulation. Furthermore under the assumption that the satellii 
attitude is stable and the movement of the sensor position is almost linear, 2D affine projection model i 
applicable to basic equations for mapping. This paper discusses the application of the 2D affine projectin 
model to high-resolution satellite imagery and SPOT scenes at various terrain area in JAPAN. The first toj 
is the approach to generate ortho-imagery using the model. The second topic is the real-time image positioniy 
on a softcopy photogrammetric workstations for satellite imagery. 
1 INTRODUCTION 
Precise digital maps generated from satellite imagery are assuming growing importance in the spatial informatio 
industry. For coverage at medium to small mapping scales, satellite line scanner imagery has a number d 
advantages over aerial photography for topographic mapping, the production of digital ortho-images and tl: 
generation of DTMs. These advantages are likely to be further enhanced with high-resolution earth observatio 
satellites like IKONOS. Notwithstanding the practical advantages of satellite imagery, however, the projection 
of satellite imagery, which is imaged with a CCD line sensor, is quite different from that of conventional aerll 
photographs. The line scanner imagery has geometry of central perspective in the scan line direction, and clo 
to a parallel projection in the flight direction. Therefore stereo scopic systems are required to special position 
modules for the satellite imagery. 
Since SPOT satellite was launched in 1986, many studies has been carried out to implement an accunit 
positioning control modules of SPOT imagery to analytical plotter. One approach of them is to incorporate tli 
inverse collinearity equations for dynamic satellite imagery in the real-time loop (Gugan, 1987). This approad 
requires high performance in CPU. Then Kratky (1989) proposed his approach which saves computing tin 
by fitting functions of inverse collinearity equations. Another approach is based on the well-known centri 
perspective model with additional parameters and uses dense look-up tables to correct the image distortiois 
(Konecny at el., 1987). Trinder’s approach (1988) is similar to it in point of using standard central perspectit 
equations, but each satellite image is divided into a number of segment 100 image wide. Small distortions 
each segment image are corrected by second order polynomial. 
In these days, some commercial softcopy photogrammetric workstations implement several satellite image 
modules with these procedures. Almost of them work effectively, but ordinarily requires high level resources ? 
computer. End users operating GIS softwares or Mapping software need lighter software which works on et 
low-end PC. This paper proposes an alternative high performance procedure for real-time posisioning contr 
of satellite imagery on digital plotting system. The approach is based on 2D affine projection model, which 
simnle forma ((Yamoto et al 1999) 
International Archives of Photogrammetry and Remote Sensing. Vol. XXXIII, Part B3. Amsterdam 2000. 
— 
2 N 
2.1 
1m } 
(5m-. 
mode 
are u 
in spi 
regar 
syste 
collin 
as th 
2.2 
The | 
exter 
numi 
paral 
low-o 
wher: 
point 
Assu 
secon 
Whes 
ally s 
y. Ne 
bet wi 
Whet 
CATT 
2.3 
In sn 
linea 
follow 
With 
expre 
Here 
chang 
algeb 
Equa 
where 
basic