Neural Networks on Handwritten Alphanumeric Character Recognition

• Main Authors: Chao Shu-Ling, 趙素玲
• Other Authors: Chou Wen-Kuang
• Format: Others
• Language: zh-TW
• Published: 1994
• Online Access: http://ndltd.ncl.edu.tw/handle/68960372869420102061

碩士 === 靜宜大學 === 管理科學研究所 === 82 === Handwritten Character recognition is a well-know complicated but intersting problem. Although numerous efforts have been made based on traditional computers, they are still suffered by eith- er time-consumed procedure or imperfect recognition rate. In th- esis, a new approach based on neural networks for recognition of unconstrained handwritten alphanumeric characters is proposed and implemented.The hybrid system consists of a kernel subsystem which is based on back- propagation networks for learning and re- cognition, and thinning. Good experimental results show the fea- sibility of the proposed approach. Since the proposed system is powerful and efficient for rec- ognition of handwritten alphanumeric characters, it has very hi- gh potential for real-time systems. In other words, it can be used to automatically read handwritten data in many forms such as tax forms, bills,and so on. On the information management po- int of view, the proposed system has achieved a significant con- tribution on man-power and time saving.