复杂背景下磨粒显微图像分割

针对DMAS系统中复杂背景下磨粒图像的自动分割问题，根据磨粒图像的相关特征，充分利用领域知识，基于图像分割和边界跟踪的思想，选取了适合磨粒图像分割的最大类间方差阈值法，并结合腐蚀、膨胀等图像处理技术，实现了复杂背景下磨粒图像的有效分割和目标提取。最后算例表明了本方法的简洁有效性。本文研究将对进一步展开磨粒识别和机械设备故障诊断研究工作提供前提和依据。

Debris Image Segmentation under Complicated Background

Aiming at the automatic segmentation problem of debris image under complicated background in DMAS, and according to the relative characteristics of debris image, and based on image segmentation and boundary tracking thinking, the Otsu method based on the maximum variance between two classes of pixels is selected to realize the segmentation of debris image by full use of the field knowledge. At the same time, the boundary tracking, erosion, and dilation techniques are combined with the Otsu method. The effective segmentation of debris image and the automatic extraction of debris objects are realized. At last, examples show that this method is able to extract effectively and automatically all the objects in debris images under complicated background. An important basis for the further study of debris recognition and machine fault diagnosis is provided.