基于零均值及零方差图构建的润滑油磨粒检测

摘要

This paper introduces a new image processing method based on zero-mean image for the wear detection of lubricating oil.Firstly, wear debris image of lubricating oil collected is processed by zero mean and zero variance. Then, the Maximum between-Cluster Variance method is proposed for segmentation of zero-mean image and zero-variance image, andthe segmentation images are fused. Finally, the opening operation and filling of morphological filtering are respectively used to eliminate noise and cavity. Experimental results show that the proposed algorithm has good ability to overcome the noise and can split the wear information completely.%提出了一种基于零均值图构建的润滑油磨粒检测的图像处理方法。首先对采集的润滑油磨粒图像进行零均值化和零方差化；然后，采用最大类间方差分割法对构建的零均值图和零方差图进行分割并对分割后的图像进行融合处理；最后，分别通过形态学滤波的开启运算和填充消除噪声和空洞。实验仿真表明，该算法具有较强的抗噪声能力，能够完整地分割出磨粒信息。