6B-3-6 
4. Character cutting and global matching to 
compute the height 
As to the geometric relationship of the 
characters and the divisions of the scale on the 
straightedge are fixed as designed value. If we 
know the distance information, the pitch size 
between every divisions of the scale could be 
computed. And then, the character can be 
approximately cut for pattern recognition, but 
with contrast modification. The procedures can 
be sum up as following steps: (1). Two 
dimensional character cutting; (2). Contrast 
modification and character combination; (3). 
Global matching with the help of character 
combination pattern and the registered 
character patterns (Figure 17); (4). Linear 
regression for height computation (Figure 18). 
We expect that the new idea will be used in the 
field of industry for device measurement and 
also for new digital level development. 
References: 
[Alberto Apostolico, Zvi Galil,1997.] 
Pattern Matching Algorithms, Oxford 
University Press, 1997 
[ Earl Gose, Richard Johnsonbaugh, 
Steve Jost, 1996], Pattern Recognition 
and Image Analysis, Prentice Hall PTR, 
1996 
Height computation: 
(Col lim ator Index) - b 
a = 
a 
(for left and right side respectively) 
5. Evaluation and discussing 
The pixel size of the CCD camera XC-75 is 
about 1 Own. the measurement distance is about 
20 meter If we use high resolution CCD 
camera such as FC1300, the pixel size is 6.7um, 
the measurement distance is expected to about 
50 meter. If we use SuperHadCCD (SONY), 
the pixel size is about 5um, so the measurement 
distance should be beyond 80 meter Of course, 
the experiments are needed, wish the success 
for the svstem in the near future