在线多笔画重复绘制草图的判定与聚类

摘要

This paper describes a method for grouping and fitting online overtraced strokes to 2D geometric primitives, which are further interpreted as 2D line drawings. Whether two strokes are overtraced is judged by the number of the sampling points of the stroke with smaller minimum bounding rectangle, which fall into the other’s tolerance zone. The grouping method does not rely on the stroke classification and can group different types of strokes together. It divides all strokes in the sketch into a set of stroke groups, which are finally fitted with geometric primitives such as straight lines, polylines and conic curves. This paper develops a hu-man-computer interface prototype system FSR-DJ using the proposed theory, which has been tested with a number of overtraced sketches and the test results show that the algorithm achieves a satisfactory interpretative efficiency.%多笔画重复绘制是设计师传递设计意图的常用方式，为在手绘图识别系统中支持多笔画重复绘制，提出一种将在线多笔画重复绘制草图解释为二维线图的判定与聚类算法。首先通过笔画逼近折线段的折点序列，构造一个带圆弧过渡的等距边线围成的多边形区域作为笔画容差带；然后利用2条笔画之中最小包络矩形面积较小的笔画的采样点落入另一条笔画的容差带的个数，进行多笔画重复绘制的快速判定；最后提出将草图分成若干个子草图的聚类算法，根据子草图中笔画的单笔画识别结果将子草图分为同类型子草图和混合类型子草图，并给出相应的子草图拟合方法。通过自主开发的 FSR_DJ 原型系统对文中算法加以验证的结果表明，该算法能有效地解决多笔画重复绘制草图的识别问题，为后期手绘图的识别研究奠定了基础。