中文摘要：

为了满足生产自动检测的迫切需求，基于机器视觉的非接触式检测技术成为自动检测发展的主流方向。以轴类零件为研究对象，在Matlab环境下开发了轴类零件外圆的视觉检测软件。该软件主要具有标定处理、批量处理及数据汇总3个功能。在标定处理过程中，提出了基于标准件的标定方法，减少了换算单位，克服了标定件和待检件难以保证相同物距带来的不便。在外圆尺寸获取中，分别采用最小二乘法和霍夫变换技术进行处理，结果表明利用霍夫变换技术获取的外圆尺寸精度更高。试验结果表明：该软件能够有效地对待检件进行检测判断，具有较高工程应用价值和生产推广价值。

英文摘要：

Nowadays, non - contact detection technology based on machine vision becomes the mainstream of automatic detection. Shaft parts were researched, and the visual detection software for their outer circle was developed in Maflab environment. There were mainly three important functions in the software, respectively, calibration processing, batch processing and data viewer. The calibration method based on standard parts was proposed in calibration processing. The exchangeable bases were reduced and inconvenience for guaranteeing the same object distance between the calibration parts and the detection parts was overcome. Least square method and Hough transform were separately used for obtaining the dimensions of outer circle. It is found that Hough transform has higher precision. The test results show that the software makes a judgment for the detection parts effectively, and has a very high value for engineering application and production promotion.