of different file formats being available in the industry. 
For image processing, the most commonly used ones are 
Binary, TIFF, PostScript, COT (Continuous Tone), and 
etc. It is prudent that the suppliers to ensure that file 
format as well as the storage media be compatible with 
the user's system to avoid unnecessary problems. 
Project Il - Color Orthophotos 
The second project calls for a production of color 
orthophoto maps of Tai-Chung Harbor in 1:10000 scale. 
In addition, line topographic maps of 1:5000 and 
1:10000 scale were also included in this project. The 
workflow is similar to the ones described above with the 
exception of mosaic and larger file sizes due to color 
input and output. Besides larger file sizes, the whole 
process tends to be very time consuming as a result of 
the 3-band (RGB) scanning as well as the four-color 
(CYMK) separations for hardcopy production. Softcopy 
mosaic was performed to achieve seamless results with 
a good success. 
On this project, aerial photos were taken at 1:20000 
scale using Kodak Aerochrome MS 2448 color aerial film. 
With respect to aerial photography, the decision to fly or 
not to fly as restricted by weather conditions was quite 
a challenge. Besides strict weather conditions typically 
needed for aerial photography, optimum timing according 
to low tidal schedule was an important consideration 
required by the contract. To meet such requirements, the 
window of opportunity for aerial photography became so 
narrow that it took approximately two (2) months before 
a successful flight was made. 
Aerial photo films were scanned at 22.5 um resolution on 
three red, green, blue (RGB) files. Scanning time and file 
sizes required for each photo became tripled the ones 
normally required for black and white. Therefore, it is 
critical that data storage be optimized for large volume 
productions. 
Aerotriangulation was performed on Intergraph IM 6887 
softcopy stereoplotter using PAT-M adjustment. A total 
of 440 photogrammetric pass points were measured and 
recorded which yielded a sigma naught of 8 um for 
horizontal block and 14 yum for vertical block, 
respectively. 
The workflow is similar to the ones required for the 
basemaps with the exception of no superimposition of 
topographic linework over the images. Unlike the 
basemaps, mosaic was needed to form a complete 
orthophoto map. In addition, the images were rectified 
individually for the respective RGB bands. Through the 
use of Map Publisher, color photo images were 
segregated in a 4-color (CYMK) separation and then 
output to a TIFF file for a filmwriter production. 
To prepare for an output plotting of color orthophotos, 
digital files in four-color-separations were saved in TIFF 
format. Through the use of Intergraph ImageStation and 
Photoshop, the operator was able to perform softcopy 
370 
mosaic and match color tones as necessary. For 
mosaicking, the orthophoto is digitally processed to 
remove seam lines between the mosaicked images. 
Mosaics with seamless results were achieved with ease 
due to today's softcopy technology. 
ACCURACY 
Editing was made during every stage of map compilation. 
A minimum of ten (10) percent of map quadrangles 
produced was field checked for accuracy with precise 
traverse survey on prominent ground features. The 
results of field check are listed in Table |. 
Measures of Precision. In accordance with the contract 
agreement, at least 90 percent of all points checked shall 
meet or exceed the following acceptance criteria. 
e. Discrepancy of coordinates of control points 
between that of the maps and field surveys shall 
not exceed 0.2 mm. 
$ Discrepancy of coordinates of prominent ground 
feature between that of the maps and field 
surveys shall not exceed 0.4 mm. 
e. Discrepancy of spot elevations between that of 
the maps and field surveys shall not exceed 1/4 
of the contour intervals. 
e. Discrepancy of contour lines between that of 
the maps and field surveys shall not exceed 1/2 
of the contour intervals. 
e In forestry area where it is normally covered 
with dense trees, interpretation of ground 
elevations is difficult. Discrepancy of contours 
between that of the maps and field surveys is in 
direct proportion to the tree heights. In this 
case, such discrepancy shall not exceed 1/2 of 
the tree heights. 
On both projects, all points being checked met or 
exceeded the guidelines stated above. As it can be seen 
from Table |, the results of field check are well within the 
allowable discrepancies. 
DATA STORAGE 
Storage of data in the day-to-day operations was on 1.3 
GB optical disks. A 22.5 um scanned image file is 
approximately 100 Mbytes for each photograph. Data 
was compressed to 1/3 of their uncompressed size using 
the JPEG (Joint Photographic Experts Group) 
compression. Further compression to smaller file sizes 
was not considered. The rule of thumb is that geometric 
distortion for compression ratios below 5 can be 
neglected. Data was uncompressed before rectification 
and then compressed again after rectification of the 
images. 
Storage of digital orthophotos for the entire island of 
Taiwan would require at least 350 optical disks. The 
same data could be stored on eighty (80) 8mm Exabyte 
tapes. For archives, the latter would be a good choice. 
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B4. Vienna 1996 
  
anc 
GIS 
Rei