DATA FUSION IN URBAN CARTOGRAPHY 
Oscar Ricardo Vergara' 
Ana Lücia Bezerra Candeias" 
Maria de Lourdes N. de O. Kurkdjian" 
“* INPE - Instituto Nacional de Pesquisas Espaciais 
C.P. 515 - 12201-970 Säo José dos Campos, SP, Brazil 
La {richard, analucia } @dpi.inpe.br 
‘lulu @1tid.inpe.br 
Commission VII, Working Group 9 
KEY WORDS: Remote Sensing, Human, Settlement, Updating, SPOT, Resolution, Urban Cartography, Data Fusion. 
ABSTRACT: 
The set of images of the various operating orbital sensors is not enough to meet all the needs of the users, as there are applications 
which require high spatial and spectral resolutions. For this reason, aiming at overcoming some original data restrictions, 
image processing is used to generate improved quality products. Data fusion is the generic name given to techniques which allow to 
integrate images of different spatial and spectral characteristics in order to obtain synthetic images which have the advantages 
presented by their components, making the extraction of information easier. 
In this paper, the processing applied to the original data (panchromatic and multispectral HRV-SPOT images) made the 
photointerpretation easier as well as the mapping of intra-urban areas, within a methodology of cartographic updating at 1:50,000 
scale. Such processing started by applying a high-pass filter to the panchromatic image, to enhance edges. Later the IHS 
transformation was used, following a methodology which intended to generate a synthetic product preserving the spectral 
characteristics of the multispectral component, without losing the spatial resolution of the filtered panchromatic image. The results 
obtained through this processing show that in the synthetic image, the intra-urban areas were considerably improved. 
1.INTRODUCTION 
resulting product was used to update the intra-urban areas of a 
The set of images of the various operating orbital sensors is topographic map at 1:50,000 scale. 
still insufficient to solve all the problems of Remote Sensing 
users, as there are applications that require high resolutions, 
spatial as well as spectral. An example of this is the urban 
areas mapping in topograpic maps updating, because the 
current cartographic norms demand greater thematic 
information than that offered by the original images obtained 
from the current sensor systems. For that reason it is necessary 
to use image processing techniques in an attempt to generate 
products that allow as much extraction of thematic information 
as possible. 
  
  
   
  
  
  
Table 1 - Original HRV - SPOT images identification 
The working area was a module of "Sáo José dos Campos" 
(Säo Paulo State, Brazil) topographic map at 1:50,000 scale 
(sheet SF-23-Y-D-II-1) published by IBGE in 1973. This area 
includes part of the city and is limited by the following 
geographic coordinates: S23°07’35” and  S23°14'06”; 
W45'52'07" and W45^50'06". 
Data fusion is the generic name given to techniques that 
integrate data sets of different spatial and spectral 
characteristics, obtained or not by different sensors, aiming at 
generating synthetic products of improved quality when 
compared with the original data separately. The images thus 
obtained have the informative content of the component 
products, profeting from the advantages presented by each one 
of them and making it possible to extract more complete and 
2. METHODOLOGY 
The processing applied to the original images aimed at 
accurate information (Chavez, 1986; Dutra et al, 1988: facilitating the photo-interpretation of the intra-urban areas 
Kurkdjian et Ii, 1989; Brum, 1989; Kurkdjian, 1990; Carper et included in the module under study. To do so, the following 
al., 1990; Candeias, 1992). methodology was applied: 
In this paper, the IHS transformation technique was used to 2-1 Filtering 
merge panchromatic and multispectral (BGR 1,2,3) HRV- N . Rio : {inal 
SPOT digital data generated by INPE and identified in Table 1. ~~ A high-pass isotropic filter was applied to the origina 
The image processing was made at SITIM (Integrated System panchromatic SPOT image which allowed to generate an edge 
of Image Treatment developped at the same Institute). The enhanced image. The filter used (Figure 2) did not produce any 
image dislocation effect or geometric alteration. 
734 
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B7. Vienna 1996 
2.2 
This 
fron 
chaı 
bett: 
In ti 
the 
mul 
SPC 
pres 
area 
spat 
filtei 
e Ir 
e Th 
eo TI 
e Th 
co 
e Th 
e Th: 
bai 
e Th 
12 
e In 
SP 
wh 
e Th 
RG 
e The 
The 
e=