基于亚像素定位和虚拟圆扫描的齿轮在线检测研究

采用图像识别技术对微小塑料齿轮进行质量检测,针对缺齿、齿歪、披峰等齿形误差的随机性,运用三点定圆心法实现塑料齿轮内圆的粗定位,试验数据说明其运行时间比传统Hough变换大大缩短,并为亚像素定位提供了原始数据;提出的降维灰度矩法与最小二乘圆法结合,实现内圆圆心的亚像素定位,同时运用3σ原则迭代法提高了亚像素定位精度;用"虚拟圆扫描法"实现大小轮齿的齿形检测。结果表明该方法可以满足齿轮的自动检测。

Research on the on-line inspection of micro plastic gear based on sub-pixel location and dummy circle scan method

The quality of micro plastic gears is inspected with the image recognition technology.Focusing on the situation that gears' defects are uncertain, the rough location of micro plastic gears' inner circular contour is achieved by the method of determining-circle-with-three-points.The experiments show that the running time by the new method is greatly reduced than that by the traditional Hough transform,and it can provide the primary dates for the sub-pixel location.Colligating the dimensionality reduction gray moment operator and the least square circle,the sub-pixel location of the inner circle centre is realized and the sub-pixel location precision is improved by applying the 3σ iteration algorithm.A dummy circle scan method is presented to realize the dentiform inspection,and the results indicate that it can meet request on the automatic inspection of micro plastic gears.