中文摘要：

为了解决传统边缘检测方法存在的边缘短小、边缘丢失和噪声敏感等问题,提出一种基于Sobel梯度模板的多阈值实时边缘检测方法。该方法在大阈值下收集用于边缘定位的标兵点,在小阈值下确定用于边缘连接的搜索区域,并结合一种连点画图游戏的思想,利用不同阈值下收集的边缘线索和信息提取最终的边缘。实验结果证明,该方法运行速度快,具有较好的实时性,检测的边缘为单像素宽度,定位精确,连续性好。

英文摘要：

To solve the problems such as short edge, edge loss and noise sensitivity that exist in some traditional edge detec- tion methods, a multi-threshold and real-time one based on Sobel gradient template is proposed in this paper. It collects pacesetters which are used for edge positioning under large threshold and determines search filed which is used for edge linking under small threshold..tt extracts final edges with these edge clues and information collected under different thresholds by the thoughts of a connect-the-dots drawing game. Experimental results show that this method runs fast and has good real-time performance, and edges detected by it equip with single pixel width, position accuracy as well as good continuity.