manner. These features maximize Radar Operator 
performance and minimize conditions that could lead 
to operator error. Table 2 lists the types of features 
which ROICATS incorporates. 
The OMI feature which is first and foremost is the 
ability to simultaneously display various types of 
information on a computer screen. Figure 3 illustrates 
the various computer screen display options available to 
the Operator. All of the display options are composed 
of four basic windows: the Static Menu, the Dynamic 
Menu, the Softkeys Menu, and the Display Area. 
The first window, called the Static Menu, is used to 
display alphanumeric information relevant to the 
images currently displayed. This information includes: 
the image file name, the aspect and depression viewing 
angles of the displayed image, the scale factor, the 
colour mode, and the display style mode. 
The second window, called the Dynamic Menu, is used 
to display all the alphanumeric information output by 
a computer-based tool (e.g. database contents and 
HELP descriptions). Another form of information that 
can be displayed in this window is the output from the 
automated ship classification algorithms (e.g. a list of 
candidate ship classes determined by the Expert 
System, the ship class output from each of the three 
automated algorithms, and the combined output from 
Dempster-Shafer algorithm (see Figure 2)). 
The third window, called the Softkeys Menu, is used to 
display all of the computer-based tools that the 
Operator can choose from. The tools are represented 
by software programmable buttons on the computer 
screen. A button is selected by moving the cursor of a 
mouse or trackball over the button and pressing the 
enter or select key on the mouse or trackball. 
The fourth window, called the Display Area, is used to 
display SPOT, RDP or STRIPMAP images, computer- 
based ship models, digitized ship photos or digitized 
maps. While the first three windows are fixed in size, 
the fourth window can be divided into one to four 
subwindows or areas, each capable of displaying an 
image. With a computer display screen size of 1280 
(wide) by 1024 (high) pixels, the Static Menu should be 
about 256 by 150 pixels, the Dynamic Menu should be 
about 256 by 724 pixels, the Softkeys Menu should be 
about 256 by 150 pixels, and the Display Menu should 
be about 1024 by 1024 pixels. 
The remaining OMI features listed in Table 2 are self- 
explanatory and will not be repeated here. Suffice it to 
say that this list of features represents a minimum set 
of mandatory features required for ROICATS. 
328 
5.0 CONCLUDING REMARKS 
This paper has presented an overview of the work on 
the classification of SAR images of ships performed at 
DREO. This work has resulted in the development of 
a Radar Operator Interactive Classification And 
Training System (ROICATS). The three main 
components of ROICATS were described. These 
components are: computer-based tools, operator- 
machine interface (OMI) features, and automated 
target classification algorithms. Four different 
classification algorithms were briefly discussed, eight 
tool groups were briefly described, and five OMI 
features were presented. 
The present development status of ROICATS can be 
summarized as follows. All OMI features, except for 
Online Guidance, have been implemented. The 
automated classification algorithms are at the following 
stages of completion: the attributes of the Expert 
System have been defined, but must still be 
implemented and tested; the Neural Network and 
Decision Theoretic methods are still being tested; and 
the Combinatorial Optimization classifier requires some 
minor changes and then further testing. The Image 
Processing, Database Manager and Overlay tools have 
not yet been developed. 
Anticipated future tasks, required to complete 
ROICATS, include: complete the implementation of all 
the computer-based tools, implement the Online 
Guidance feature, and finish testing all the automated 
algorithms. One other major component which will 
eventually be part of ROICATS is a Computer-Based 
Training subsystem. This subsystem will be used to 
familiarize Radar Operators with the methodology of 
classifying radar images of ships. It is hoped that the 
first three main components of ROICATS will be 
completed within one year, and the training subsystem 
within two years. 
ACKNOWLEDGEMENTS 
This work was supported by the Research and 
Development Branch of the Department of National 
Defence. The review and editing of this paper by 
Anthony Damini is very much appreciated. 
REFERENCES 
[1] Blackman, S.S., 1986. Multiple-Radar Tracking 
with Radar Applications, Artech House. 
[2] Rey, M.T. and Vant, M.R., 1987. "Image 
Enhancement and Display Techniques Applied to 
SAR580 Images of Ships," Communications Research 
Centre Report No. 1421, Ottawa, April 1987. 
  
  
Figure 1: 
  
LIBE 
  
SH] 
  
INPUT 
SPOT/I 
IMAGE 
A SHII