The Expanded PDCS Digital Image Processing System (EPDCS) 
by 
G. H. Xu 
Chinese Academy of Forestry 
Wan Shou Shan, Peking, Chin& 
Abstract 
This paper is a research report describing the Computer Soft- 
ware System EPDCS, developed for processing remote sensing imagery 
data. The programme system has grown out of the PDCS programmes 
which the author developed and implemented in Sweden. 
The system includes the following groups of programmes: 
-Tape handling and subscene extraction 
-Statistics and histognomne print 
-Image enhancement 
-TIraining area extraction,calculation and display 
-Supervised classification 
-Unsupervised classification 
-Cleaning-up of a classified map 
-Àceuracy evaluation 
-Result output 
In this paper, discussions are made about the techniques for 
the preprocessing, enhancement and classification of MSS data, with 
greater elaboration on the classification. 
The system has been used in land use mapping and environmental 
monitoring. The results have shown some promise of usability in 
practical work and will be reported in another paper. 
INTRODUCTION 
One application of remote sensing is the use of Multi Spectral 
Scanner (MSS) date for land use inventory over larg areas. This 
report discusses the Expanded PDCS programme system developed for 
processing remctely sensed MSS data by the computer systems, both 
at the Department of Physical Geography, University of Stockholm 
and at the Chinese Academy of Forestry. 
The programme system consists of a series of computer routines 
for image data handling, manipulation and classification, with the 
following two points considered in the course of the development of 
the programme package: : 
1. The programme system should be suitable for research work. 
2. It should be possible for a user to run the programmes even 
without prior experience of advanced computer programmes. 
Proceeding from these points, on the one hand, the EPDCS system 
was designed to derive different methods for dealing with the same 
subject. For instance, there are several methods for an image en- 
hancement ,tbraîning area extraction,supervised classification and 
NU classification. That is useful either for selecting 
optimum method or for determining suitable condition of the diffe- 
rent method, On the other hand, the programme system was written