Abstract: For locating circle objects detection which appears commonly in machine vision, an algorithm is put forward, which combines Blob analysis with least-square-fitting to detect circle objects. Blob analysis is used to locate circle objects roughly, which gives out target circle center and radius as the results. According to the output, boundary points are detected and used as the input of least square circle fitting. Detecting target circles and locating the middle of the circles automatically are carried out by this algorithm, and the experiment has been carried out on vision system based on DSP using piano string poles as an example. The results show that it can meet the real-time requirement and detect the object accurately.

Key words: machine vision, location, Blob analysis, least-square-fitting, Digital Signal Processor（DSP）

摘要： 针对视觉处理中常见的圆形目标检测，提出了结合Blob分析与最小二乘法圆拟合的定位检测方法，利用Blob分析算法实现粗定位，根据粗定位的输出检测目标圆的边界点，将其作为输入参数给最小二乘法圆拟合进行精定位；实现了在视场范围内对圆形区域自动检测及中心定位，在DSP视觉系统上以钢琴弦柱加工为检测背景对算法进行验证，实验结果表明算法满足工业应用中的实时性和精确性要求。

关键词: 机器视觉, 定位, Blob分析, 最小二乘法, 数字信号处理器（DSP）