摘要
提出一种手绘圆的快速识别算法,该算法采用像素跟踪方式,通过提取和跟踪圆的骨架像素得到圆片段的直线段序列,然后进行直线段序列的分解和合并处理,把属于同一圆的直线段序列连接起来,进而计算圆的几何参数,完成手绘圆的识别.与Hough变换方法相比,该算法不需要大量的计算时间和存储空间.实验结果表明,该算法能快速、准确地识别出手绘图形中的圆,具有良好的稳定性和鲁棒性.
A fast algorithm for freehand circle recognition from images is proposed in this paper. It employs the way of pixel tracing. Firstly, the circle skeleton pixels are extracted, from which the line segments of freehand circle are obtained by tracing pixels. Furthermore, the line segments that belong to the same circle are grouped together through the processing of line segments. Finally, we analysis the grouped line segments and get the geometrical parameters of extracted circles. Unlike Hough Transform based algorithm, this method expends low costs in computation time and storage space. The experiment results demonstrate the freehand circles can be recognized quickly and accurately, and the proposed method is stable and effective.
出处
《延边大学学报（自然科学版）》
CAS
2008年第4期283-286,共4页
Journal of Yanbian University（Natural Science Edition）
基金
延边大学研究生科研项目(延大科研合字(2008)第30号)
关键词
手绘图形
手绘圆
识别
直线段
freehand drawings
freehand circle
recognition
line segments
