基于改进PCNN的汽车油封尺寸检测方法

Method for Detecting Size of Automobile Oil Seal Based on Improved PCNN

针对由于汽车油封固有的弹性给人工尺寸检测精度带来很大不确定性的现象,为满足于当前高效快速的现代化生产和汽车油封的高精度尺寸检测要求,提出一种汽车油封高精度尺寸检测方法。以PCNN模型为基础,为保留油封的真实边缘,将双边滤波和Canny算子的思想与PCNN算法相结合,利用Visual Studio 2015软件平台对油封图像进行边缘检测,获得清晰且连续的理想边缘图像。采用随机霍夫变换圆检测的优化算法对油封边缘点进行圆拟合,完成油封内径和外径的尺寸检测。将视觉检测结果与三坐标测量结果进行对比,结果证明,基于改进PCNN的汽车油封尺寸检测结果与三坐标测量结果的平均误差为0.0176 mm,单个油封检测时间为2.17 s,满足油封尺寸高精度要求。由此说明,提出的基于改进PCNN的汽车油封尺寸检测方法是有效可行的。

In view of the fact that the inherent elasticity of automobile oil seals brings great uncertainty to the accuracy of manual size detection,in order to meet the requirements of current efficient and rapid modern production and high-precision size detection of automobile oil seals,a high-precision size detection of automobile oil seals method is proposed.In this pa-per,based on the PCNN model,in order to retain the true edge of the oil seal,the idea of bilateral filtering and Canny op-erator is combined with the PCNN algorithm;and the edge detection of the oil seal image is performed by using the Visual Studio 2015 software platform to obtain a clear and continuous ideal edge image.The optimization algorithm of random Hough transform circle detection is used to fit the edge points of the oil seal to complete the size detection of the inner di-ameter and outer diameter of the oil seal.Comparing the visual inspection results with the three-coordinate measurement results,it is proved that the average error between the inspection results of the automotive oil seal size based on the im-proved PCNN and the three-coordinate measurement result is 0.0176 mm;and the detection time of a single oil seal is 2.17 s,which meets the high precision requirements of the oil seal size.The result show that the improved PCNN-based car oil seal size detection method in this paper is effective and feasible.