Fast algorithm for character line extraction from handwritten examination papers
Katsunari Goto, Chinthaka Premachandra, Shinji Tsuruoka, Haruhiko Takase, Hiroharu Kawanaka · 2014
Academic institutions such as universities and technical colleges usually conduct paper-based examinations and reports for evaluating students' academic performance. Teachers must take considerable time for paper marking in the paper-based examinations. To reduce the paper marking burden of the teachers, we are developing an automatic paper marking system. To execute paper marking, handwritten character lines are extracted from examination papers, and then characters on those lines are recognized. In this paper, we mainly discuss about the character line extraction from handwritten examination papers without ruled line. The extraction of character lines from the non-ruled papers is difficult through image processing due to the writing characteristic of the students. Furthermore, extraction accuracy becomes an important factor for the character recognition performance. The conventional character line extraction algorithms for printed document does not show better performance for this problem. At the beginning of this study, we propose a method combining two conventional methods. But, it took a dozens of seconds to implementation. Thus, in this paper we develop a new method having better performance in character line extraction and processing time reduction.