Photoelasticity Color Dynamics Dataset for stress analysis applications
收藏资源简介：
Photoelasticity is an optical technique that takes advantage of birefringence in some materials to obtain isoclinic and isochromatic fringe patterns. Which represent the direction and the difference of the principal stresses in a sample under a mechanical load. The changes in the photoelasticity fringe patterns contain information about the stresses presented in a sample. One of the ways to access this information without unwrapping the fringe patterns is through pixel intensity values. Which allow to reconstruct temporary signals that represent the changes of stresses produced by external stimulations such as mechanical loads or temperature changes. Today there is no dataset in the literature that contains data abourt this type of information. To obtain this data set, sequences of 720 images of synthetic photoelasticity models with different geometries were generated and a progressive mechanical load between 0 and 4000 Newtons was applied. The data set presented below contains two arrays called testing and training which have the following dimensions [8000,2161] and [32000,2161], respectively. The rows represent the number of pixels from which information was obteined and the columns are divided into 4 sections. The columns between 1 and 720 correspond to the RED channel, the columns between 721 and 1440 correspond to the GREEN channel and data in columns 1441 and 2160 correspond to the BLUE channel. Finally, the data in column 2161 corresponds to the classification of each pixel. The classification is divided into 4 labels. Number 1 corresponds to pixels located in high stress zone, number 2 corresponds to pixels located in a medium stress zone, number 3 corresponds to pixels located in low stress zone, and number 4 corresponds to pixels located in a zone with the presence of residual stresses. The stress classification criterion was based on the fringe order, orders greater than 6.5 were taken as high stresses, between 2 and 6.5 as medium stresses, and below 2 orders were considered as low stresses. The category of residual stresses was taken based on a geometric selection of an area with the presence of residual stresses.
光弹性法（Photoelasticity）是一种光学测试技术，其借助部分材料的双折射（birefringence）效应，可获取等倾线（isoclinic）与等色线（isochromatic）条纹图案。此类图案能够表征机械载荷作用下试样内主应力（principal stresses）的方向与幅值差值。 光弹性条纹图案的演化蕴含着试样内部应力分布的相关信息。无需对条纹图案进行相位解包裹（unwrapping the fringe patterns）即可提取该信息的主流方法之一，便是利用像素强度值（pixel intensity values）。通过像素强度值，可以重建出表征外部激励（如机械载荷或温度变化）所引发应力变化的暂态信号。目前，学界尚未有公开数据集涵盖此类数据。 为构建该数据集，研究人员生成了包含720帧图像的序列数据集，对应不同几何构型的合成光弹性模型（synthetic photoelasticity models），并对模型施加了0至4000牛的递进式机械载荷。 本数据集包含测试集（testing）与训练集（training）两个数组，其维度分别为[8000, 2161]与[32000, 2161]。数组的行代表提取了有效信息的像素总数，列则被划分为四个部分：第1至720列对应红色（RED）通道，第721至1440列对应绿色（GREEN）通道，第1441至2160列对应蓝色（BLUE）通道。 最后，第2161列的数据为每个像素的分类标签。该分类共设有四个类别：编号1对应高应力区域像素，编号2对应中应力区域像素，编号3对应低应力区域像素，编号4对应存在残余应力（residual stresses）区域的像素。 应力分类标准基于条纹级数（fringe order）制定：条纹级数大于6.5的被划定为高应力区域，2至6.5之间的为中应力区域，小于2的则归类为低应力区域。残余应力类别则通过对存在残余应力的区域进行几何选区的方式确定。
