|
KMID : 0380319930520010041
|
|
Journal of Korean Research Institute for Better Living
1993 Volume.52 No. 1 p.41 ~ p.54
|
|
|
Recognition of Hangul-Character using Skeleton from the Bending Points
|
|
|
|
|
Abstract
|
|
|
Abstract-This paper proposes a new tracing technique about Hangul character recognition through the direction code and the variable tracing window. To obtain features in the given character image, the corresponding window is defined firstly by
the
size
of image. And then triangulation of those characters at bending points are obtained during the scanning process. The generated features are 'ㄱ'bending -point, 'T' bending-point and ending-point. For the higher recognition rate, we need more
features
such as line directions and combined directions. All the line directions are generated from mid points on the triangle mesh. Triangle mesh is automatically constructed by using the bending-points on character boundary. The conceptual simplicity
in
finding skeletons reduces the scanning time. And Hangul characters are classified based on the types of bending-points and their direction attributes. Experimental results shows that the presegmented Hangul characters are well classified by
proposed
algorithm.
|
|
|
KEYWORD
|
|
|
|
|
|
FullTexts / Linksout information
|
|
|
|
|
|
Listed journal information
|
|
|
|