Figure 4 Extraction of feature point by means of image processing 
Figure 5 Extraction of radiation color 
P2-3-2 
Figure 3 Configuration of ergoma system 
Recorder) so that PC can continuously capture a scene of 
sequential images. TV monitor displays sequential images played 
back by LVR. 
3 ERGOMA SYSTEM 
3.1 Feature Extraction 
Feature area has been extracted using color information by the 
authors. However, it was ascertained that opening in 
morphological image processing has prominent ability to extract 
a feature area [Yamada and Chikatsu, 1999]. Therefore, opening 
was adopted for extracting a feature area in this paper. 
Figure 4(a) shows the original image under the insufficient image 
quality and figure 4(b) shows the extracted maker area by 
opening. On the contrary, figure 4 (c) shows the extracted maker 
(a) Original image (b) Extracted marker area by 
opening 
(c) Extracted maker area by old (d) Final marker area in old 
method method 
area by old method using color information and figure 4(d) shows 
the final results that noises were eliminated in figure 4(c). From 
these figures, It is concluded that opening is efficient method for 
extraction of a feature area. 
On the other hand, distinction between welding and burnishing 
could be achieved utilizing the difference of radiation color 
between the two. For example, radiation color for welding gives 
RGB (205,205,255) and burnishing gives (205,205,195). Then, 
distinction between the two could be achieved using the 
difference of the RGB values. Figure 5(b) and (d) shows the 
extracted radiation area for welding (figure 5(a)) and burnishing 
(figure 5(c)). 
Furthermore, these distinctions are expressed on the PC monitor 
such as "welding" or "burnishing" and these informations are 
recorded as a part of database for ergoma system. 
(a) Welding image (b) Extracted welding radiation 
(c) Burnishing image (d) Extracted burnishing radiation