内螺纹图像识别技术研究

Research on Image Recognition of Internal Thread

针对生产中内螺纹难以快速自动检测问题,提出了一种基于计算机视觉技术的内螺纹非接触式自动检测方法。采用数学形态学处理二值化后的内螺纹图像,消除了内螺纹小径圆弧边缘缺口、裂缝及破洞等缺陷;然后应用最小二乘圆拟合方法检测内螺纹小径圆弧,得到精确的小径半径尺寸,以内螺纹小径的半径尺寸差异为判别条件,开发自动检测系统,实现内螺纹的准确识别。实验结果表明,该系统检测精度高、正确率高、速度快,能满足内螺纹生产自动检测的要求。

A non- contact automatic detection approach from computer vision technology is proposed to address the challenging issue of rapid and automatic detection for internal thread screw in production. Defects like notch,flaw,and breakage on the small diameter edge of internal thread screw are removed via the image binarization with mathematical morphology. Then,precise value of the small radius is detected with use of the least square circle fitting technique. An automatic detection system was developed with small radius as the criterion,which can accurately identify the internal thread screw. The experimental results show that the system can meet the requirements on the shop floors due to its nature of high precision,correct rate,and speed.