连续运动螺纹尺寸自适应机器视觉检测

Continuous Motion Thread Size Detection with Adaptive Machine Vision

针对螺纹零件尺寸检测效率低、工作强度大、检测精度不高的问题,提出了一种连续运动螺纹尺寸自适应机器视觉的检测方法。首先设计了一套在线图像采集装置用于采集运动螺纹零件的图像;其次使用NCC(normalized cross correlation)归一化匹配算法,完成对螺纹零件的识别及追踪,在此基础上设计了自适应的ROI(region of interest)裁剪区域;然后通过Canny算法提取边缘轮廓,利用最小二乘法直线拟合技术对螺纹轮廓进行直线拟合,生成螺纹中径线,完成对螺纹中径的测量;最后在使用Harris角点检测算法提取多螺纹轮廓波峰与波谷的基础上,提出了利用螺纹中径线与各角点之间的距离来完成对角点的细化。实验表明,所提方法在螺纹尺寸检测方面具有可行性和有效性。

Because of the problem of low efficiency,high working intensity and low detection accuracy of threaded parts,the adaptive machine vision detection method of continuous motion thread size is proposed.Firstly,an online image acquisition device is designed to collect moving thread parts images.Secondly,NCC(normalized cross correlation)normalization matching algorithm is used to identify and track the threaded parts.On this basis,an adaptive ROI(region of interest)clipping region is designed;then the edge contour is extracted by Canny algorithm.The least squares straight line fitting technique is used to fit the thread profile straight line,the thread diameter line is generated,and the measurement of the thread diameter is completed.Finally,based on the Harris corner detection algorithm is used for multi-thread contour peak and trough extraction,it is proposed to use the distance between the thread diameter line and each corner point to complete the refinement of the diagonal point.Experiments show that the proposed method is feasible and effective in thread size detection.