玻璃小球直径的图像测量方法

提出了一种基于数字图像处理技术的微小玻璃球直径的测量方法。该方法采用Sobel算子计算目标的梯度图像，取阈值对其进行二值化处理；根据边缘成像特点，通过数学规划法计算小球边缘的初始位置；在此基础上，采用空间矩亚像素定位算法对目标边缘进行精确定位；通过拟合这些边缘点，即可求得目标直径。实验结果表明，该方法对二值化阈值的依赖性小，抗噪声能力强：在一定的范围内，阈值的不同对该算法重复精度的影响小于0．06个像素；由于采用交互方式选定处理区域，算法的计算速度较快，完成一次计算的时间不到1S。

Image measurement for the diameters of micro glass spheres

For measuring diameters of micro glass spheres, a method based on digital image processing is presented. Sobel algorithm is used to calculate gray-level gradient image and then transform it to binary image with the threshold. According to the imaging property of edges, the primary locations of object edge are calculated by means of mathematical programming. And then the spatial moment operator is used to calculate the accurate locations of object edges. By fitting these edge points, accurate diameter can be acquired. Results of experiment show that the method is independent of binary threshold and robust to additional noises. Effect of threshold variations within a finite scope on repeat accuracy is less than 0.06 pixels. Due to manual selection of processing region, calculation process is fast and the computing time is less than Is.