MI-Cored：一种基于机器视觉的米字形结构元素数学形态学滤波方法 MI-Cored: A Mathematical Morphological Filtering Method by Using Structural Element of “M” Glyph Based on Machine Vision

随着信息化技术的不断发展，机器视觉技术受到了大众的广泛关注。机器视觉技术具有精确、高效、快速、灵活等优点，使其在工业生产中具有巨大的应用潜力和效能。本文针对目前国内道岔钢轨尺寸传统测量方法的弊端：效率低、精度低、人为主观性强和测量者个体差异等问题，提出了一种基于机器视觉的道岔钢轨端面尺寸测量方法，此方法的核心技术在于应用米字形结构元素的数学形态学操作MI-cored对道岔钢轨端面图像进行二次去噪处理，并通过实验验证，应用MI-cored滤波方法后最终得到图像的信噪比最小，具有最佳的噪声点抑制效果。通过此方法得到最优图像供道岔钢轨端面尺寸测量，提高尺寸测量的准确性，进而保障道岔钢轨生产质量。 With the development of information technology, Machine vision technology which has the ad-vantages of accurate, efficient, fast and flexible, receives extensive attention of the public. The application of machine vision technology in industrial production has great potential and effectiveness. Due to the manual measurement of the turnout rail size has problems such as low efficiency and accuracy, subjective and poor consistency, this paper proposes a new mathematical morphological filtering method based on machine vision, called MI-cored which uses a structural element of “M” glyph. Through experiments, by using the MI-cored, the resulting image signal has the smallest noise-signal ratio and the best effect of noise suppression, which is provided to measure the turnout rail size, improves the accuracy of measurement and ensures the quality of turnout rail production.