微纳尺寸零件曲率半径测量方法

Radius of curvature measurement for micro-nano components

为精确检测非透明微纳尺寸零件的曲率半径,提出一种以零件的带标尺数字图像为分析对象,对提取的零件轮廓边缘进行最小二乘曲线拟合和曲率半径计算的方法。采用改进的中值滤波算法和直方图均衡化预处理微型金刚石颗粒的扫描电镜图像,通过Canny算子检测到零件轮廓边缘后,利用高阶非线性多项式函数拟合轮廓采样坐标点,并结合图像度量标尺得到了零件的实际曲率半径。误差分析结果表明,在观测倍率为1 200、轮廓采样点数量为80、拟合曲线阶次为20时,新方法对金刚石颗粒的曲率半径测量误差不超过10%,具有较高的测量精度和工程实用价值。

To acquire the radius of curvature （ ROC） for opaque micro-nano components with high precision, a new technology was introduced in the research, according to the digital image with measuring scale of the investigated subject, and ROC was calculated after the contour edge detection and curve fitting based on the least square law. The modified median filtering was presented to pre-process scanning electron microscope （ SEM） images of the ex-perimental diamond particle, together with the histogram equalization method. Then, the nonlinear polynomial with high order was used to fit the sampling points on the contour edge of the diamond particle detected by the Canny edge detection operator. Further, the actual ROC of the components was calculated combining with the measuring scale on SEM image. The measuring error analysis shows that the deviation of the actual value using this technology is less than 10% for the diamond particle, and as the magnification rate of SEM, the amount of sampling points on contour line and the order of fitting model are 1 200, 80 and 20 respectively, which indicates a high measuring preci-sion and application value.