欢迎您！东篱公司
退出
-
申报数据库
-
立项数据库
-
成果数据库
-
项目获奖数据库
中文摘要：
基于Curvelet变换理论，提出了利用多尺度多重分形谱参数表征发动机典型磨粒图像特征的新方法。利用Curvelet变换对磨粒图像进行分解与单尺度重构，获得不同尺度的磨粒子图像。运用多重分形理论对磨粒子图像进行研究，首先计算子图像的多重分形谱，得到磨粒的多尺度多重分形谱；然后提取磨粒的多尺度多重分形谱参数，并探讨多尺度多重分形谱参数与磨粒形态特征的内在联系。将多尺度多重分形谱参数用于典型磨粒识别，识别成功率达到90％。研究结果表明：多尺度多重分形谱能全面反映发动机磨粒的形态特征，不同类型磨粒的多尺度多重分形谱参数有较大差异，可以作为磨粒的新的有效特征参数。
英文摘要：
To better express morphology characteristics of wear particles, a novel method utilizing multiscale multifractal spectrum parameters to describe features of wear particle images was proposed. Wear particle subimages in different scales by Curvelet-transform-based decomposition and single-scale reconstruction were studied with the multifractal theory. Multifractal spectrums of every subimage were calculated firstly and defined as multiscale multifractal spectrums; then multiscale multifractal spectrum parameters of the wear particles were extracted, and correlation between multiseale multifraetal spectrum parameters and characteristics of wear particle morphology was examined. The success rate of recognizing typical wear particles by using the multiscale multifraetal spectrum parameters can reach 90~. This shows that the multiseale multifractal spectrum can depict morphology characteristics of wear particles. Different wear particles have obviously distinguishable multiscale multifractal spectrum parameters, they are a kind of novel effective feature parameters of the wear particles.
同期刊论文项目
同项目期刊论文
期刊信息
位置：