원문정보
초록
영어
This paper describes a system to classify the off-line handwritten Hindi characters into several groups based on some similarity measure. A novel method is proposed for finding the header line, based on end points and pixels positions in the top half part of the character image. The algorithm works in the presence of slant of the header line. After the identification and removal of header line, all the characters are coarse classified. A new algorithm is designed for the identification of presence and position of vertical bar in the handwritten Hindi characters. A coarse classification rate of 97.25% has been achieved in the simulation study.
목차
Abstract
1. Introduction
2. Preprocessing
2.1 Database Collection
2.2 Identification and Removal of Header Line
3. Coarse Classification
3.1 Identification of Non-Connected Characters
3.2 Identification of End-bar, Middle-bar and Without-bar Characters
3.3 Identification of Closed loop
4. Experimental Results
5. Conclusions
References
1. Introduction
2. Preprocessing
2.1 Database Collection
2.2 Identification and Removal of Header Line
3. Coarse Classification
3.1 Identification of Non-Connected Characters
3.2 Identification of End-bar, Middle-bar and Without-bar Characters
3.3 Identification of Closed loop
4. Experimental Results
5. Conclusions
References
저자정보
참고문헌
자료제공 : 네이버학술정보