融合分割与智能聚类的铁谱图像处理及其评价体系的研究

项目名称： 融合分割与智能聚类的铁谱图像处理及其评价体系的研究

项目编号： No.51205202

项目类型： 青年科学基金项目

立项/批准年度： 2013

项目学科： 机械工程学科

项目作者： 王静秋

作者单位： 南京航空航天大学

项目金额： 25万元

中文摘要： 本项目以提高铁谱图像处理与分析的智能化和自动化水平为目标，力图解决磨粒分割这个制约铁谱图像处理与分析的难点。拟通过选择适于铁谱图像处理的颜色空间并优化颜色分量，寻求并行的智能化图像分割与聚类算法，处理具有多特征、宽尺度、模糊边界的目标分割与聚类问题，以实现磨粒与背景、磨粒与磨粒之间的快速准确分割。并且通过建立合理的评价体系，实现分割聚类算法及参数的优化。本项目将为后续的磨粒群体特征提取和磨损状态识别打下良好的基础，对于提高我国航空发动机和大型低速重载设备的工况监测和故障诊断水平具有理论意义和实用价值。

中文关键词： 铁谱分析；图像分割；磨粒分割；分割评价；磨粒识别

英文摘要： This research is trying to resolve the difficult problem, the segmentation and clustering of the wear debris in ferrography image, so as to improve the level of intelligent and automatic analysis for ferrography image. Firstly, different types of color space will be studied and compared to seek the optimization of the color components which are suitable for the processing of ferrography image. Secondly, in order to achieve accurate segmentation between wear particles and background, and between particles deposited in particle chains, the parallel and intelligent algorithms will be studied to deal with the segmentation and clustering of the object region in ferrography image, which have the properties of multiple features, wide size scale and fuzzy boundary. Finally, a reasonable evaluation system will be established to evaluate the process and results of segmentation and clustering to optimize the algorithms and parameters used. This project will establish a solid foundation for the subsequent wear debris feature extraction and wear state recognition. It is of important theoretical significance and practical value for improving intelligent ferrography analysis and fault diagnosis for aeronautical engines and equipments working in low-speed and high-load conditions.

英文关键词： Ferrography；Image segmentation；Wear particle segmentation；Segmentation evaluation；Wear particle identification