【作者】 黄艳艳；

【导师】 曾致远；

【作者基本信息】 华中科技大学 ， 系统分析与集成， 2007， 硕士

【摘要】 粗粒土料是土石坝的主要填料,高土石坝的流变已成为坝工专家和工程师广为关注的问题,其流变特性是研究的基础。粗粒土料的流变特性是进行土石坝流变研究的基础,但目前对粗粒土料流变特性的试验研究欠缺。由于粗粒土的离散特征,粗粒土的很多力学特性都难以用现有的连续体力学理论予以反映,作为一种散体材料,土颗粒间位置排列和粒间作用力对于粗粒土的力学性质有重要的影响。本文首次从图像分析的角度,利用MRI(Magnetic resonance imaging,核磁共振成像)技术获得粗粒土组构图像,采用颗粒图像的分割和特征提取方法分析观测粗粒土试样受力变形过程中内部结构的动态变化,探索建立其力学体系的方法和途径。粗粒土试样的纵向断面图很好地展现了压力试验下石料颗粒的状况。对粗粒土图像进行研究的一个关键步骤是通过图像分割算法,将每一个颗粒作为目标提取出来,分析其形态信息,并进一步得出颗粒之间相互作用的情况。针对图像的特点,本文先对图像做最优阈值分割,得到二值图像,白色部分为颗粒,黑色部分为背景。最优阈值分割所确定的颗粒边缘轮廓线一般很准确,但可以发现大部分颗粒之间“粘连”在一起。针对颗粒粘连部分轮廓线曲率较大的特点,可以通过求角点的办法先大致定位。本文选取了Harris角点检测算法,该算法稳定性好,对噪声不敏感,对角点的检测准确性较高。最后在粗粒土组构图像测量系统中实现了组构要素的定量统计分析,为后续的组构力学效应分析和建立组构力学模型提供了很好的数据支持。更多还原

【Abstract】 Granular material is the main material of embankment dams and is one of the major study objectives of soil mechanics. The rheology problem of high earth rockfill dam has been widely contemplated by experts and engineers specialized in dam engineering,and the rheology property of coarse stuff adopted as the stuff material is the base of research. But so far, studies in this area are relatively few. As a kind of coarse aggregate,it was shown that the arrangement and intergranular action of particles had important impact on the mechanical property of granular material and many problems related to the granular material fabric.The principal achievements of the fabric research of granular material were presented.This dissertation firstly raises a point that from the perspective of image analysis, through MRI(Magnetic resonance imaging) technology,to observe the dynamic change of granular material internal structure during the process of force and deformation.The Longitudinal section image of granular material shows the situation under pressure well.Segmentation is a key step during the particle image processing.The segmentation result has direct influence on the counting accuracy and the following processing. Each particle will be extracted as a goal for form information analysis,and further,the situation of the interaction between the particles will be studied.According to the characteristics of images, the optimal thresholding method is used to get binary image. The boundary of particles is generally accurately determined.But we can find some particles tend to cluster into each other.So According to larger curvature of Crosspoint,Harris corner detector is selected to separate the clustering particles because the algorithm is stable, not sensitive to noise.Finally Motion Image of Granular Material Fabric Measurement System is achieved to analyse quantitatively the element of the fabric and provide a good support for mechanical effects analysis of fabric and the establishment of mechanical model.更多还原