马尔可夫随机场在纸病检测中的应用研究

Research on the Application of Markov Random Field in Paper Disease Detection

关键词:  纸病图像  MarKov随机场  纹理特征参数  最大差

Key Words:paper disease image  MarKov random field  texture feature parameter  maximum difference

摘要:纸病检测是造纸生产过程中重要的环节，现有的纸病检测系统一般采用阈值算法或边缘检测算法对图像进行分割。为解决阈值分割和边缘检测分割方式中存在的误分以及过度分割问题，本研究提出了基于马尔可夫（MarKov）随机场的纸病图像分割方法。通过MarKov随机场理论对纸病图像纹理进行分析得到纹理特征参数，利用纹理特征参数以及最大差值对正常背景和纸病区域进行分割。结果表明，相比于其他分割算法，基于MarKov随机场的纸病图像分割方法可有效提取出纸病图像的纹理细节和轮廓特征，提高分割的准确度。

Abstract:Paper disease detection is an important part of paper production process.Image segmentation is a necessary step in paper disease detection.In order to solve the problems of error segmentation and over-segmentation in threshold segmentation and edge detection segmentation， this paper proposed disease image segmentation method based on MarKov random field.The texture feature parameters were obtained by using MarKov random field theory to analyze the paper disease image texture.The normal background and the paper disease region were segmented using the texture feature parameters and the maximum difference.Experiment results showed that compared with other segmentation algorithms， the Markov random field paper image segmentation method effectively extracted the texture details and contour features of the paper disease image， and improvd the segmentation accuracy.