基于知识的窗口矢量化技术

Technique of Knowledge-Based Window Vectorization

介绍了一种工程图扫描识别中基于知识库推理的自适应窗口矢量化技术.它既避免了常用的基于细化的识别处理方法在交叉点处引起的畸变,又较好地实现了交叉点的精确定位.文中还给出了自适应窗口的约束条件和黑游程结构图的构造方法,同时详尽地分析了知识库的组织机制.

Vectorization is a key issue in the process of recreating engineering drawings from scanned input image. This paper introduces the technique of adaptive window vectorization based on knowledge base inference. It not only avoids distortion caused by the commonly used thinning based recognition process at cross point, but also achieves the accurate positioning of cross point. The constraint conditions of adaptive window and the approach of constructing black run length structure drawing are also given. Besides,the organization mechanism of knowledge base is analysed in detail.