Abstract:
This paper presents a new method to obtain the complete strokes based on the abstract of dynamic structure which is of value in theory. proposes the numeral recognition system which is made up by advanced ART1 model and BP models. Experiments show that the proposed approach is promising.
摘要: 针对手写体数字的特点并从实用性的角度出发,提出了基于动态特征抽取完整笔划的方法,具有一定的理论价值。介绍了一种由改进的ART1模型和BP模型串联而成的数字识别系统。通过实验,表明数字识别效果良好。
关键词:
数字识别,
笔划延续最大优势方向,
定向搜索邻域,
ART,
BP神经网
CLC Number:
FAN Yanfeng1,3, XIAO Le3, ZHEN Tong2,3. Research in Handwriting Digital Recignition[J]. , 2005, 31(10): 177-179.
范艳峰,肖乐,甄彤. 自由手写体数字识别技术研究[J]. , 2005, 31(10): 177-179.