currently(D. MARR, 1982). Moreover, the filter size in this The al 
case is a minimum size of 5X5 for the LG filter because of includir 
being taken up to a fine point of the edge. However, since cultural 
even up to noise are detected as edges in the edged artifact, 
images obtained by using the filter. It is difficult to judge pattern 
which part is a true edge and how is the shape of pattern image" 
or feature even if taking a look at only the edged image conclu: 
directly. include 
Thus, i 
by us 
charac 
given it 
manpo 
Theref 
employ 
makinc 
than th 
  
  
  
  
  
  
  
  
The ed 
proces 
(1) det 
of wh: 
connec 
feature 
Mr. Su 
the fea 
compo 
the € 
(T.SUC 
This he 
3.1 EDGE DETECTION IN FILTERING METHOD of brigl 
AND PROBLEMATICAL POINTS E RTL T REPRE TA SETI “un 200 
  
ES 
= 
ERE 
Figure 1(b) Target 2(central projection image) 
  
3. EDGE DETECTION METHOD 
  
the re 
influen 
The h 
obtain: 
image 
variatic 
Moreo 
weight 
a size 
calcule 
An edge detection images are important data for straw- 
rope pattern pottery(e.g. archeological artifacts at Jomon 
period) because of being used when an archaeologist 
does a chronological classification referring to the direction 
of bonds attached to the surface. The authors, with the 
purpose of reducing a tracing time have been developing 
so far with the automatic making of edged image as one 
function of the system by using a filtering technology. 
Figure 2 shows edged images obtained relative to the 
images in Figure 1. Table 1 shows a parameter of 
Laplasian Gausian Filter(size:5 x 5) 
h(x, 
Table 1 Parameter of Laplasian Gausian Filter(size: 5 x 5) 
  
Where 
  
-1 -5 -7 -5 -1 
-5 -9 8 -9 -5 
-7 8 78 8 -7 
-5 -9 8 -9 -5 
-1 -5 -7 -5 -1 
  
h,(x, 
EU 
  
  
  
  
  
  
  
  
  
  
  
These edged images have been made by a  Laplasian à; rS : Na : - | | 
Gausian (LG) filter which is known as a filter having no 
Fi 2(b di Target.2,LG filter 
directionality among many filtering techniques available garezipj Edgedaimage(Targ cha; ier) 
666