边缘图像中圆形轮廓的两步法检测  

Two-Step Method for Detection of Circle Profiles in Edge Images

作　　者：于中[1] 徐文立[1] 陈峰[1] 

机构地区：[1]清华大学自动化系,北京100084

出　　处：《计算机辅助设计与图形学学报》2005年第2期243-246,共4页Journal of Computer-Aided Design & Computer Graphics

基　　金：国家"十五"重点科技攻关项目 (2 0 0 1BA60 9A)

摘　　要：针对边缘图像中的圆形可能局部畸变的情况 ,提出了利用边缘分段拟合进行圆形检测的两步法 详细描述了该算法的具体步骤 ,并与随机Hough变换 (RHT)进行了对比实验 实验结果表明 ,文中算法鲁棒性强 。Considering that the circular edges in binary edge images might be partly damaged, a two-step method of arc fitting using segmented edges is proposed to detect circles. In the paper, the algorithm is described in detail, and comparison experiments with the randomized Hough transformation (RHT) have been made. The test experiment shows that the proposed method is robust, efficient and practical.

关 键 词：圆形检测 拟合 图像识别 算法 

分 类 号：TP391.72[自动化与计算机技术—计算机应用技术]