KEY WORDS: 
ABSTRACT 
COMPRESSION OF REMOTELY SENSED DATA USING JPEG 
Dafer Ali Algarni 
Assistant Professor 
King Saud University, Riyadh, Saudi Arabia 
Commission III Working Group 4 
Image, Compression, Automation, Classification, Digital, Distortion 
The automation complexity may be lessened by compressing digital images without effecting the 
image fidelity. The Joint Photographic Experts Group (JPEG) algorithm is tested for compressing 
remotely sensed data, e. g. Landsat TM images. The compressed images are compared to the 
original images at different rates of compression in two different experiements. It is found that 
JPEG can be as useful in digital mapping as it is in video and other visual applications. 
Compressing a complicated scene to about 12%, which saves more than 700,000 bytes of three TM 
band with a size of 786,432 bytes, is possible with irregular degrading in the visual quality. Beyond 
this limit, the image is highly degraded. This may meet certain mapping applications where other 
measures, rather than high accuracy, are sought. The statistical analysis shows, however, that JPEG 
can not be recommended for precise mapping, and that the geometric and visual quality of the 
output of compression is a scene-dependent matter and can not easily be generalized for all images. 
1. INTRODUCTION 
The increase in the methods of data-gathering 
and automation for the purpose of digital mapping 
are strongly inter-related. For automation to be 
well accomplished, it needs the contribution of 
different information and techniques. On the other 
hand, too much information may cripple the 
advance of automation, making the process of 
obtaining useful information very slow and 
confusing. 
Furthermore, if we were to process different data 
with different 
characteristics, 
difficult. 
numerical and symbolic 
automation will become very 
In discussing the implementation of softcopy 
workstations, Miller et al. (1992) stated that ‘‘the 
generic problems are in the image processing 
area....While storage of such images is no longer 
a serious problem, fast accessing and processing 
certainly are’’. Therefore, the amount of data and 
24 
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B3. Vienna 1996 
qualiy have a direct effect on the amount and 
quality of automation. If the processed data is 
somehow reduced to a little but sufficient and 
correct amount, not only the image processing 
is facilitated but also the possibility of 
automation will increase and many other related 
mapping problems will be solved. 
The main theme of this paper is to study the 
applicability of the JPEG technique for 
compressing remotely sensed data. This 
technique was found to be useful in reducing 
and transmission of still images for visual 
applications (Paik, 1992). A study was made on 
compressing aerial image of smooth distinctive 
features using JPEG shows that a 10% reduction 
can be used without degrading the visual or 
geometric quality (Lammi, and Sarjakoski, 
1995). It is of great interest to test the 
applicability of JPEG to remote sensing images 
of subtle texutre varations. In this study, a 
simple measure of evaluation is used where the 
  
  
  
  
   
  
  
  
   
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
   
  
  
  
  
  
  
  
  
   
   
  
  
  
   
  
   
  
   
   
    
    
effect of 
such as i 
is organ 
next sect 
evaluate 
section 
effect o 
experim 
four, an 
JPEG is 
image c 
stored d 
such d 
compres 
those of 
et. al 1 
redundai 
and Mi 
basically 
domain 
main pr 
namely 
Figure 1 
INPUT 
DC’ 
IMAGE 
Geel JDC] 
CONSTRACTE 
IMAGE 
Figure 
In the e 
through 
function 
can be 
follows