首页> 外文会议>Air & Waste Management Association's annual conference & exhibition;A&WMA's annual conference & exhibition >Extracting an Optical Finger-Print – a New Approach toSingle Particle Analysis
【24h】

Extracting an Optical Finger-Print – a New Approach toSingle Particle Analysis

机译:提取光学指纹–单颗粒分析的新方法

获取原文

摘要

A fully automated system has been developed for microscope-based single particle analysisby extracting optical finger prints from individual particles in ambient air samples. For thispurpose, light microscopy was utilized to facilitate an objective measuring technique byemploying a novel pattern recognition technique. This automated particle classification isbased on so-called grey scale invariants, extracted from microscopic images of translucent,fluorescent and dark field microscopy. This information was bundled to a feature vectorproviding a kind of finger print for every particle. In a first step this approach was used for anautomated recognition of allergen carriers such as pollen and fungal spores. A leave-one-outtest gave a recognition rate of about 95% for 26 of the most frequent pollen species in centralEurope. Because no pollen-specific code was used, the recognition software was alsoemployed, without changes, for an automated recognition of fungal spores. Six of the mostfrequent airborne fungal spore genera in Central Europe were classified with a meanrecognition rate of 93%. These results provided the basis for a research project aiming at thedevelopment of a fully automated system. The instrument should combine (1) high-volume sampling of coarse particles >2.5 μm, (2) electrostatic precipitation of this fraction onto asurface suitable for optical analysis, (3) automatic preparation for microscopic single particleanalysis, (4) imaging by various microscopic techniques, e.g. transmitted, fluorescence anddark field microscopy, (5) feature extraction by grey scale invariants, (6) classification byself-learning Support Vector Machines and (7) hourly output of number concentration ofairborne pollen, fungal spores and other particles of interest. A prototype was demonstrated inearly 2005. First field tests are planned for the first half of 2005. A commercialised deviceshould be available as of 2007. The project is funded by the German Ministry of Educationand Research.
机译:通过从周围空气样本中的单个颗粒中提取光学指纹,已经开发出了一种用于基于显微镜的单颗粒分析的全自动系统。为了这个目的,利用光学显微镜通过采用新颖的模式识别技术来促进客观测量技术。这种自动的粒子分类基于所谓的灰度不变式,该灰度不变式是从半透明,荧光和暗场显微镜的显微图像中提取的。该信息被捆绑到一个特征向量上,为每个粒子提供一种指纹。第一步,此方法用于自动识别过敏原载体,例如花粉和真菌孢子。一次留守测试对欧洲中部最常见的26种花粉物种的识别率约为95%。由于未使用花粉特定的代码,因此无需更改即可使用识别软件来自动识别真菌孢子。对中欧最常见的空气传播真菌孢子属中的六个进行了分类,平均识别率为93%。这些结果为旨在开发全自动系统的研究项目提供了基础。仪器应结合(1)大量采样> 2.5μm的粗颗粒;(2)将该部分静电沉淀到适合光学分析的表面上;(3)自动准备进行微观单颗粒分析;(4)通过各种显微镜成像技术,例如透射,荧光和暗场显微镜检查;(5)通过灰度不变式提取特征;(6)通过自学习支持向量机进行分类;(7)每小时输出空气中花粉,真菌孢子和其他感兴趣颗粒的浓度浓度。原型于2005年初进行了演示。计划于2005年上半年进行首次现场测试。商业化的设备将于2007年推出。该项目由德国教育和研究部资助。

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