【摘要】 细化在图像识别中有非常重要的地位，基于细化的识别处理算法是工程图自动识别处理中使用较为广泛的一种技术，然而，目前的细化过程会在节点处引入畸变而产生较多的细碎矢量，使得系统对基本图元，即直线和圆弧的识别率较低。本文对细化后畸变节点形态进行了深入地分析，提出了一种基于自适应视口的节点校正算法，有效地抑制了细化后节点的畸变，从而使得后续的跟踪及识别更为准确，该算法已成功地用于解决实际问题。更多还原

【Abstract】 ? Thinning occupies an important place in image recognition. Thinning based recognition algorithms are widely used in the field of automatic processing of engineering drawings, However, the current thinning processes perform with too many left over broken vectors which result in relatively low recognition rate of the primitive graphic elements such as lines and arcs(including circles). The node patterns after thinning are deeply analyzed, and a new node regulation algorithm in the self zooming viewport which can effectively revise the deformation of node created in the thinning procedure has been proposed. The following trace and recognition procedure can provide more exact results on the basis of the regulation step. The algorithm has been applied to solve the practical problem successfully. 更多还原