• 主题
高级搜索  >
历史搜索 清空历史
首页> 外文期刊>Nordic Pulp & Paper Research Journal >Spectral analysis and classification of dirt particles in pulp
【24h】

Spectral analysis and classification of dirt particles in pulp

机译：纸浆中灰尘颗粒的光谱分析和分类

获取原文
获取原文并翻译 | 示例
           
页面导航
  • 摘要
  • 著录项
  • 相似文献
  • 相关主题

摘要

The presence of dirt particles in the pulp affects the quality of the final product. Thus the cleanliness of the pulp plays an important role in defining the quality of the paper. In the pulp and paper industry, the presence of dirt particles is often evaluated by visual inspection. The results of visual inspection vary from person to person. Hence different imaging techniques have been suggested for online and offline inspection of the pulp. In this study, we measured the spectral reflectance of pulp sheet from visible to near infrared region and we applied principal component analysis (PCA) method to analyze spectral images of the pulp sheets with dirt particles. Furthermore, the segmentation of the dirt particles, dirt particle counting and classification of the dirt particles were studied. Three thresholding algorithms were compared for the segmentation of dirt particles in the pulp sheets. Dirt particle counting results were compared with a commercial dirt particle counting system called the Digital Optical Measurement and Analysis System (DOMAS). Our proposed method gives better results than the DOMAS in segmentation and in dirt particle counting, because the spectral images has more information than low contrast grayscale image imaged in the DOMAS. In addition, dirt particle classification was studied using two different algorithms; the linear discriminate analysis (LDA) and the single layer feedforward network (SLFFN). PCA subspace feature of dirt particles was used for classification of dirt particles. We obtained classification rate greater than 95% and kappa statistics greater than 0.9 with less than eight dimensions of PCA subspace. In addition, based on the PCA results of dirt particles, we propose optical dirt particle filters to classify dirt particles. These optical filters provide good results on differentiating dirt particles.
机译：纸浆中污垢颗粒的存在会影响最终产品的质量。因此，纸浆的清洁度在确定纸质方面起着重要作用。在纸浆和造纸工业中，通常通过目视检查来评估污垢颗粒的存在。视觉检查的结果因人而异。因此，已经提出了不同的成像技术用于纸浆的在线和离线检查。在这项研究中，我们测量了纸浆板从可见光到近红外区域的光谱反射率，并应用主成分分析（PCA）方法分析了带有污垢颗粒的纸浆板的光谱图像。此外，研究了污物颗粒的分割，污物颗粒计数和污物颗粒的分类。比较了三种阈值算法对纸浆片中污垢颗粒的分割。将污垢颗粒计数结果与称为数字光学测量和分析系统（DOMAS）的商业污垢颗粒计数系统进行了比较。我们提出的方法在分割和污垢颗粒计数方面比DOMAS提供更好的结果，因为光谱图像比DOMAS中成像的低对比度灰度图像具有更多的信息。此外，还使用两种不同的算法研究了污垢颗粒的分类；线性判别分析（LDA）和单层前馈网络（SLFFN）。尘埃颗粒的PCA子空间特征用于尘埃颗粒的分类。在PCA子空间少于8个维度的情况下，我们获得的分类率大于95％，kappa统计值大于0.9。此外，基于污垢颗粒的PCA结果，我们提出了光学污垢颗粒过滤器以对污垢颗粒进行分类。这些滤光片在区分灰尘颗粒方面提供了良好的效果。

著录项

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

客服邮箱：kefu@zhangqiaokeyan.com

京公网安备：11010802029741号 ICP备案号：京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有

  • 客服微信

  • 服务号