类椭圆特征自动识别及亚像素提取的完整实现  

Complete approach to automatic identification and sub-pixel center location for ellipse feature

作　　者：薛婷[1] 孙梅[2] 张涛[1] 吴斌[3] 

机构地区：[1]天津大学电气与自动化工程学院,天津300072 [2]天津内燃机研究所,天津300072 [3]天津大学精密测试技术及仪器国家重点实验室,天津300072

出　　处：《光电子．激光》2008年第8期1076-1078,共3页Journal of Optoelectronics·Laser

基　　金：国家自然科学基金资助项目(50705065);天津市自然科学基金资助项目(08JCYBJC10000)

摘　　要：针对视觉检测中高精度、自动化类椭圆图像特征提取的需要,提出一种类椭圆特征自动识别和中心亚像素定位的完整实现方法。方法中采用基于类椭圆边缘属性对特征区域进行自动识别,采用最小二乘椭圆拟合精确求取类椭圆亚像素定位中心。实验表明,基于该解决方案,可以实现类椭圆特征的自动识别及椭圆中心的亚像素定位,定位算法精度较高,鲁棒性强,同时简便易行,不需要人机交互,可以很好地满足视觉检测仪器化需求。To meet the need of automatic image features extraction with high precision in visual inspection, a complete approach to achieve automatic identification and sub-pixel center location for similar-ellipse feature is proposed with correlative key technologies. In the method,the automatic identification of feature area is fulfilled based on the edge attribnte, and the sukrpixel center location is accomplished with the leasw-square approach. The experimental result show that automatic identi fication and higher precision sub-pixel center location for similapellipse feature can be achieved with this mthod. Meanwhile this method can meet the need of instrumentation of visual inspection because of easy realization and without man-machine interaction.

关 键 词：类椭圆特征 自动识别 亚像素 边缘属性 

分 类 号：TN247[电子电信—物理电子学]