基于直线段跟踪的手绘图元离线识别方法

摘要

手绘图形是人类思维外化和表达意图的一种有效方式,如何有效地提取手绘在图纸上的图形元素是理解绘图者意图的关键问题.鉴于手绘图形是由基本图元组合构成,采用层次结构逐步实现图元提取的思想,提出了一种手绘基本图元(线段、弧、圆和椭圆)的离线识别方法.在提取图形笔画骨架像素的基础上,跟踪骨架像素得到图形的直线段描述;通过对直线段序列的分析,进行直线段序列的断开和连接处理,形成图元的曲线段描述,通过对图元曲线段描述的分析得出图元的几何参数.实验表明,该方法能够以高精确度快速识别出图像中包含的手绘图元,具有良好的稳定性和鲁棒性.%Freehand shape is an efficient way for conveying and expressing human ideas. The key point for understanding the drawers' ideas is extracting the graphics primitives drawn on paper. By employing a hierarchical approach to extract the graphics primitives from images, this paper proposed an off-line recognition method for graphics primitives (line segment, arc, circle and ellipse). Firstly, described the image as a set of midline pixels, from which extracted line segments, and furthermore, broke line segments up and formed curve segments by linking line segments. Finally, calculated the geometrical parameters of extracted graphics primitives. The experiment results show that the proposed method is stable and effective, and the graphics primitives are recognized from images with high speed and high accuracy.