基于离散曲率特征的弧线形状检测方法  

Arc contour detection method based on discrete curvature characteristics

作　　者：张旭东[1] 赵其杰[1] 

机构地区：[1]上海大学机电工程与自动化学院,上海200444

出　　处：《上海大学学报（自然科学版）》2017年第5期702-713,共12页Journal of Shanghai University:Natural Science Edition

基　　金：国家自然科学基金资助项目(61101177)

摘　　要：物体形状检测是实现机器人自主环境理解的基础.针对机器人作业场景经过色彩分层及多尺度滤波分割后的物体表现为连通域,以及便于分析形状的特点,提出了一种基于离散曲率特征的物体轮廓弧线形状检测方法.该方法将物体轮廓提取、直线和特征点检测作为基础,剔除影响弧线检测的轮廓直线,并根据剩余轮廓各点处的离散曲率滑动变化特性检测弧线特征.对机器人作业场景实物进行实验,弧线形状检测的平均正确率达到90.6%,处理时间为0.75 s,表明该方法能有效地对物体轮廓弧线形状进行检测.Detection of object shape is the basis of autonomous environment understanding for robots. An object turns into connected domains after color layering and multi-scale filtering segmentation, and therefore it can be easily analyzed. In view of this, a method based on discrete curvature characteristics is proposed to detect the arc shape of objects. This method is based on extraction of object contour, and detection of lines and features. Interferential lines are illuminated, and the arc features are detected according to the discrete curvature changes of the remaining contour points. A system of robot operation is established for experiments. Average precision of arc contour detection is 90.6% and the handling time is 0.75 s. The result shows effectiveness of the proposed method in detecting arc contour of objects.

关 键 词：弧线检测 物体形状 服务机器人 环境理解 

分 类 号：TP242[自动化与计算机技术—检测技术与自动化装置]