INTERPRETABILITY OF GEOGRAPHIC INFORMATION FROM HIGH RESOLUTION SATELLITE IMAGES 
Toshiaki Hashimoto 
CEReS(Center for Environmental Remote Sensing), Chiba University 
Japan 
E-mail: hashi@rsirc.cr.chiba-u.ac.jp 
Commission II, Working Group 1 
KEYWORDS: Interpretability, High resolution images, Geographic information 
ABSTRACT 
One of the trends of sensor development is the improvement of spatial resolution. It will be expected that more information will be 
extracted from the images of higher resolution. In this study, the interpretability of geographic information from high resolution images 
are examined. The examination was performed using simulation images generated from aerial photos. Some conditions of real satellite 
images like spatial resolution, atmospheric effects, dynamic range, etc. are taken into consideration in the generation of simulation 
images. The geographic information to be interpreted are those containing in the existing topographic or thematic maps such as roads, 
railways, buildings, water bodies, forests, etc. Through the examination, it was expected that most of geographic information could be 
extracted from the images of the same resolution as ALOS satellite or better. 
1. INTRODUCTION 
This study was conducted on the following background. 
(1) The launch of ADEOS/AVNIR, the plan of ALOS/PRISM 
[RESTEC, 1995], AVNIR-2 and the development of 
commercial high resolution satellites. 
(2) The needs of revising technique of GIS data in proper cost. 
The object of this study is to evaluate what kinds of geographic 
information can be interpreted from high resolution satellite data. 
2. METHODOLOGY 
2.1 Test Area 
Two kinds of test areas were selected in Japan. One contains 
urban areas, agricultural areas, forest areas. Another contains 
mountainous regions. 
2.2 Simulated Images 
Figure 1 shows the procedure for the generation of simulated 
images. 
(1) The simulated images were generated from color air photos of 
1/10,000. The photos were digitized into 1 meter resolution. 
(2) To consider the atmospheric effects, the digitized images were 
blurred by the Point Spread Function with clear atmosphere 
[Kawata, 1997]. 
(3) The spatial resolutions of simulated images were determined 
under the assumption that ALOS, ADEOS/ AVNIR and 
commercial satellites would be utilized. 
(4) The dynamic range of digitized images were different from 
that of real satellite images. The former was adjusted to the 
dynamic range of ADEOS/AVNIR images. 
(5) The high-resolution data will be compressed onboard for 
downlinking and decompressed on the ground. The same 
process was adopted for the simulation. 
(6) The Pa/Mu composite, or the Pan-sharpened image, were used 
for interpretation to take both advantages of the high 
resolution of panchromatic band and the spectral character 
istics of multi-bands. 
Figure 1 Procedure of simulation image generation 
PI-4-1