基于SEM图像的阈值回归法计算印刷纸张孔隙率

A Porosity Calculation Method for SEM Image of Paper Based on Threshold Regression

关键词:  回归模型  纸张孔隙率  图像处理  SEM图像

Key Words:regression model  paper porosity  image processing  SEM image

摘要:为提高基于扫描电子显微镜（SEM）图像处理的印刷纸张孔隙率计算方法的准确率和适应能力，本文提出了一种基于阈值回归的纸张孔隙率计算方法，并建立SEM图像分析程序。本方法首先针对小部分样品同时进行物理实验和SEM图像处理，将两部分数据组合到一起并用于训练阈值回归模型获取模型参数；然后针对其他样品，通过回归模型获取图像分割的最佳阈值，并应用于基于SEM图像的非涂布纸张孔隙率计算过程。验证实验表明，该阈值回归模型应用于非涂布纸张孔隙率计算时，能有效提高阈值选择的客观性和计算结果的稳定性，同时提高了计算机对非涂布纸张SEM图像处理方法的适应能力。

Abstract:To improve the accuracy and adaptability of paper porosity calculation method based on scanning electron microscope (SEM) image processing, a method for calculating porosity of paper images based on threshold regression was proposed. The method firstly carried out physical testing and image processing simultaneously for a small number of paper samples，and combined the data from physical testing and image processing to train the threshold regression model and obtain the model parameters. Then, the optimal threshold of image segmentation was obtained through regression model and applied to the calculation process of paper porosity for uncoated paper samples based on its SEM images. The verification experiments showed that this method could objectify the selection process of segmentation threshold in the image processing process. The objectified segmentation threshold could effectively reduce the result error and make the method have higher adaptability in various types of paper.