【摘要】 基于单义域邻接图的扫描图样矢量化,要生成具有较强整体性的条形域,包括线段、圆弧、箭头和多义线。它们的分类特征和分类知识均有模糊性,采用模糊推理进行分类。在分类时,以条形域较小和较大的直线度、圆形度和等宽度为模糊集合,以计算较少的升降半梯形为相应的隶属函数。这样分类符合人类认知方法,而且算法简单易于实现。更多还原

【Abstract】 The stripes with globality are important for vectorization of scanned drawings based on primitive region adja-cency graphs.The stripes consist of lines,arcs,arrows and polylines.Features of stripes and their knowledge for classifica-tion are fuzzy.This paper presents a method of classification of stripes using fuzzy reasoning.Smaller and larger linear,circular and regularity of width are fuzzy sets,and their membership functions are ascending and descending halving trapezoids with smaller computation.The method is simply and robust.更多还原