稀疏轮廓关联的直线整体矢量化算法

直线是工程图样矢量化过程中重要的识别对象,它的准确矢量化对图样的识别效果起到重要作用。基于光栅直线几何特征的局部和整体信息之间的关联,提出了稀疏轮廓关联的直线整体矢量化算法。该算法通过稀疏轮廓跟踪搜索直线的特征区域,得到直线特征区域的几何属性,利用获得的直线特征区域的几何属性指导直线后续的延拓跟踪。在延拓跟踪过程中,利用跟踪到的稀疏轮廓信息对直线进行延拓并进行校正,获得精确的矢量化结果。该算法在时间和空间上具有一定的优势,对直线的识别具有完整性和任意性、识别率高、具有强抗噪性等特点。

A Global Vectorization Algorithm for Sparse Contour Associated with Line-typed Graphic Primitives

Line vectorization is an important processing step in graphics recognition procedure.It plays a critical role in vectorizing the lines of engineering drawings accurately and efficiently.Based on the local and global association of line geometric features,a global vectorization algorithm of sparse contour associated with graphic primitives based on sparse contour tracking is put forward.With this algorithm we can obtain the geometric attribution of characteristic segment by tracking the contour of raster graphics primitive. The next step of tracking is guided by the attribution of characteristic segment in order to get the accurate vectorization result of lines.The advantage of the proposed algorithm is its high recognition rate.Experimental result shows that the line vectorization algorithm has good anti-noise ability and is significantly improved compared with the existing vectorization algorithms.