【摘要】 为了实现渗碳齿轮内氧化自动测量评级，课题组提出了一种基于图像处理的内氧化最大深度测量方法，并设计了一种内氧化定量分析软件。首先，对采集的金相图进行预处理，降低金相图的噪声并去除点状形态的内氧化组织；其次，针对光学显微镜成像过程中边缘模糊问题，提出了一种基于灰度变化率的K均值算法，改进后的算法能够有效地分割内氧化组织与背景区域；最后，利用亚像素对内氧化组织边缘进行计算，提高了最大深度测量的精确度。对15组齿轮内氧化金相图进行了测量实验，结果表明误差均在2%以内，验证了测量算法的有效性和可靠性。该系统实现了内氧化自动评级功能。更多还原

【Abstract】 In order to realize the automatic measurement and rating of internal oxidation of carburized gears, a maximum depth measurement method of internal oxidation based on image processing was proposed, and a quantitative analysis software of internal oxidation was designed. Firstly, the collected metallographic diagram was pretreated to reduce the noise of the metallographic diagram and remove the point-shaped internal oxidation tissue. Secondly, aiming at the edge blur problem in the imaging process of optical microscope, a K-means algorithm based on gray change rate was proposed. The improved algorithm can effectively segment the internal oxidized tissue and the background region. Finally, the sub-pixel was used to calculate the edge of internal oxidized tissue, which improved the accuracy of maximum depth measurement. The measurement experiments were conducted on 15 sets of internal oxidation metallographic diagrams of gears, and the results show that the errors were all within 2%, which verifies the effectiveness and reliability of the measurement algorithm. The system realizes the automatic rating function of internal oxidation.更多还原