摘要
根据工程图纸中字符的特点，定义了字符轮廓的长宽比、形状复杂度、内孔数和区域密集度，阐述了基于这些轮廓特征的字符分离方法，且将神经网络应用于工程图纸的字符识别，取得了满意的识别效果。
According to the characteristics of characters in engineering drawings,the length to width ratio,the shape complexity and the area density of character contour are defined.A method for characters separation based on these contour characteristics is presented.Neural networks are introduced to the recognition of characters in engineering drawings,and the recognition results are satisfied.
出处
《深圳大学学报（理工版）》
EI
CAS
1997年第4期19-24,共6页
Journal of Shenzhen University(Science and Engineering)
