高等别线纹尺中心线提取算法的研究

Study on Algorithm for Center Line Extraction of High-Grade Line Scale

摘要:

受制造技术、磨损以及污染等因素影响,对线纹尺大批量自动化检定技术的实现存在较大困难；在线纹尺特征的基础上,针对边缘对比度高但噪声严重的线纹尺,先边缘初定位将图像处理区域限制在线纹边缘区域,再边缘精定位,寻找边缘区域灰度变化最大的点,剔除不连续的边缘点后拟合边缘线,实现线纹中心线的提取；针对自身变形导致成像模糊的线纹尺,背景区域与线纹区域分别求取各列灰度均值,再利用最小二乘拟合直线求交点实现线纹中心线提取；对高等别三等标准线纹米尺多次检定,结果表明,同一刻线最大差值为1.4 μm,标准差为0.66 μm,提出的算法鲁棒性好,满足高等别线纹尺的检定要求。

Abstract:

Affected by manufacturing technology, wear and pollution, the automatic calibration technology of line scale is limited to some extent. Based-on the characteristic of line scale, and for the line scales which the edge is high-contrast with noise badly, the algorithm limit the image processing region to the edge region firstly, and then search for the dots with maximum variation of gray level, discontinuous edge points are kicked, at last the edge lines are fitted and the center line is extracted. For the line scales which edge is blurred because of distortion, the algorithm fitting two straight lines of background and line regions, and get the crossover point, finally the center line is extracted. The third grade line meter scale is calibrated automatically more than once, and the maximum difference of the same line is 1.4μm, the standard deviation is 0.66μm, the algorithm is fit for the calibration of high-grade line scale and the robustness is good.