铁谱磨粒图像多焦点融合算法研究

铁谱分析技术是一种常用的磨损监测技术。受限于高倍物镜下的景深限制,一张铁谱大磨粒图像往往只有局部聚焦清晰的特征。为了能够解决在高倍物镜下铁谱大磨粒图像的自动化清晰采集以及高质量图像融合问题,设计并构建一套自动化扫描显微系统,该系统可进行多焦点铁谱图像的自动扫描采集;同时,提出一种基于相位一致性的铁谱磨粒图像多焦点融合算法,对自动扫描的多焦点图像进行融合,得到清晰的磨粒图像。实验结果表明,设计的自动化扫描显微系统能快速完成多焦点铁谱图像的自动化采集流程,提出的图像融合算法相较于传统的小波图像融合算法具有更高的图像评价质量,并能获得更加清晰的图像边缘信息。

基于HOG和SVM算法的磨粒图像在线监测技术

为了解决发动机润滑油液磨粒图像监测只适用于微流且易受气泡干扰等问题,设计了一种可适用于相对大流量工作环境的油液磨粒光学图像在线监测系统,区分气泡和磨粒。通过该监测系统,批量采集了一系列磨损颗粒和气泡图片,用于后续图像分类算法的训练与测试。采用了一种基于背景差分和大津法的运动物体提取算法提取出大量磨粒及气泡图像样本,运用基于方向梯度直方图(Histogram of oriented gradients,HOG)进行特征提取和支持向量机(Support vector machine,SVM)分类算法对气泡和磨粒进行识别。实验结果表明,该监测系统能有效采集磨粒及气泡图像并进行自动识别。与基于形态学特征提取算法以及K最近邻(K⁃nearest neighbor,KNN)等传统分类算法相比,HOG⁃SVM算法分类精度更高,识别准确率可达83.8%。

Nanostructural origins of irreversible deformation in bone revealed by an in situ atomic force microscopy study

The structural origins of bone toughness at the nanoscale are not completely understood.Therefore,we performed in situ scanning using atomic force microscopy during macroscopic mechanical testing of antler and bovine bone,to reveal the origins of the irreversible plastic deformation at the mineralized collagen fibril(MCF)array and MCF levels.We found that the plastic deformation behavior at the nanoscale level could be divided into two stages.The first stage of plastic deformation at the nanoscale level was characterized by slippage between the MCF arrays,which contained mineral aggregate grains with regular shapes under load.In the second stage of nanoscale plastic deformation,the MCFs broke through the bonds of the extrafibrillar mineral aggregate grains and exhibited interfibrillar slippage.These nanoscale plastic deformation behaviors may thus be the origins of stress whitening and irreversible plastic deformation.Thus,the findings in this study not only shed light on the plastic deformation mechanisms of MCF arrays and MCFs,but also provide structural and mechanistic insights into bioinspired materials design and mechanisms of relevant bone diseases.