ORTHOIMAGE GENERATION IN A GIS ENVIRONMENT 
Tian- Yuan Shih 
Associate Professor 
Department of Civil Engineering 
National Chiao-Tung University 
Tainan, R.O.C. 
Eugene E. Derenyi 
Professor 
Department of Surveying Engineering 
University of New Brunswick 
Fredericton, N.B. Canada 
ISPRS COMMISSION IV 
ABSTRACT 
An orthoimage is the most practical format for utilizing digital images for map revision and resource mapping. Accordingly, 
orthoimage generation has been implemented in a geographic information system. This module includes the ground control point 
selection, image transformation, and mosaicking procedures. Both the functional design and the algorithm selection issues are 
discussed. The utilization of digital orthoimages for map revision is illustrated on an example. 
KEY WORDS: Orthoimage, Rectification, Mosaicking, Map revision, Document scanners, GIS. 
1. INTRODUCTION 
With the advent of digital, soft copy, photogrammetry, 
interest in orthoimages is rapidly increasing. An orthoimage 
is a very versatile product. It can serve as a general purpose 
digital base map, like its hard-copy predecessor, the 
orthophoto map. A companion or backdrop to a digital line 
map and a base for resource inventories are other useful 
applications. In fact, the inventory information itself can be 
derived directly from an orthoimage by visual interpretation 
or digital image analysis. Map revision is another potential 
use. 
Digital orthoimage formation is based on the same theory of 
differential rectification as its hard-copy counterpart. There 
are, however, variations in the production environment. 
There are several high precision, dedicated 
hardware/software installations in existence. Systems such 
as PRI2SM from International Imaging Systems (Boniface, 
1992), the image station module MGE from Integraph and 
the USGS system described in Skalet, et al. (1992), clearly 
belong to this category. Software packages, which utilize 
general purpose computers for orthoimage production, are 
also available. ORTHOMAP from Galileo Siscan and the 
ERDAS Digital Orthomodule are notable examples. Each 
production environment has its pros and cons. 
The dedicated, stand-alone systems employ high resolution 
precision scanners for analogue to digital conversion and 
generate high quality products which meet class "A" map 
accuracy standards. The generation of digital elevation 
models by digital image correction may also be part of the 
process. Needless to say that such installations are 
expensive and only affordable by large organizations. 
Markets for such high precision products are presently 
limited. There is, however, a demand for less expensive, 
medium precision digital orthoimages in the resource 
mapping field. 
One alternative for satisfying this need is to use a geographic 
information systems (GIS) workstation for the production 
and storage of digital orthoimages. GISs are becoming the 
norm for storing resource inventories and digital base 
mapsand a multi-function GIS would serve the needs of 
small and medium size organizations well. The Computer 
Aided Resource Information System with Raster Image 
Extension (CARI/RIX) is such a facility. 
221 
CARIS is a GIS marketed by Universal Systems Ltd. of 
Fredericton, N.B., Canada, while RIX was developed at the 
University of New Brunswick [Derenyi, 1991]. RIX 
supports the superimposition of vector data on a raster 
image backdrop, on-screen digitization, image classification, 
analytical photogrammetric operations and a variety of image 
processing and geometric registration tasks in interactive and 
batch mode. 
2. THE PRODUCTION PROCESS 
The discussion of orthoimage production in CARIS/RIX is 
limited to the use of aerial photographs, although the 
capability exists for handling satellite data as well. The main 
steps of the process are: 
* analogue to digital conversion of the photograph in a 
scanner digitizer, 
ground control point (GCP) acquisition, 
interior and exterior orientation of the photographs, 
differential rectification and resampling and 
mosaicking. 
e e e e 
2.1 The Scanning Process 
The objective was to devise a low cost scheme which suits 
the needs of resource mapping and map revision. Therefore, 
a desk-top document scanner was chosen for the digitization. 
In recent years, significant improvements were made by 
manufacturers in the performance of these scanners. 
Although these low cost models are primarily intended for 
desk-top publishing, Drummond and Rosma (1989) proved 
their potential for cartographic applications. Document 
scanners typically have a 300 to 400 dots per inch (0.085 to 
0.061 mm) geometric resolution, eight bits per pixel (256 
gray levels) radiometric resolution and a scanning surface of 
8.5 by 14 inches (216 by 355 mm). Unfortunately this 
surface is slightly narrower than the 230 mm standard size of 
aerial photographs and one of the fiducial marks and a 
narrow strip of the photo may be lost in the digitization. 
Scanners with 11 by 17 inches (279 by 432 mm) digitizing 
surface are, however, readily available at somewhat higher 
cost. 
Data output from most scanners is in TIFF format while the 
Integrated Pixel Value (IPV) format is used for raster image 
files in CARIS. The program REFOTIFF (REFOrmating 
from and to TIFF) handles this incompatibility. 
| 
M 
| 
— 
mc SPCR SE SS