应用化学 ›› 2022, Vol. 39 ›› Issue (1): 196-204.DOI: 10.19894/j.issn.1000-0518.210459

• 研究论文 • 上一篇    

仿生视觉色度识别的浓度快速测定

刘志浩, 谭晓晴, 梁永鹏, 夏超, 孟建新(), 李风煜()   

  1. 暨南大学化学与材料学院,广东省功能配位超分子材料及应用重点实验室,广州 510632
  • 收稿日期:2021-09-09 接受日期:2021-11-11 出版日期:2022-01-01 发布日期:2022-01-10
  • 通讯作者: 孟建新,李风煜
  • 基金资助:
    国家自然科学基金(21874056);国家重点研发计划(2016YFC1100502);暨南大学广东省功能配位超分子材料及应用重点实验室资助

Optesthesia Inspired Chroma Analysis for Rapid Chromatic Concentration Determination

LIU Zhi-Hao,TAN Xiao-Qing,LIANG Yong-Peng,XIA Chao,MENG Jian-Xin(),LI Feng-Yu()   

  1. College of Chemistry and Materials Science,Guangdong Provincial Key Laboratory of Functional Supramolecular Coordination Materials and Applications,Jinan University,Guangzhou 510632,China
  • Received:2021-09-09 Accepted:2021-11-11 Published:2022-01-01 Online:2022-01-10
  • Contact: Jian-Xin MENG,Feng-Yu LI
  • About author:llifengyu@jnu.edu.cntmjx@jnu.edu.cn
  • Supported by:
    the National Natural Science Foundation of China(21874056);the National Key Research and Development Program(2016YFC1100502);Guangdong Provincial Key Laboratory of Functional Supramolecular Coordination Materials and Applications

摘要:

对分光光度计检测设备的依赖,以及通过系列浓度样品测量建立标准曲线的操作限制,极大限制了分光光度法使用场景。 当前日益增长的日常健康监测、大样本量的临床测试、野外环境监测等巨大需求,渴望通过触手可及的设备,实现快速、便捷的样本定量分析。受视觉对色度快速识别的启发,设计了一种深度学习辅助的比色法,利用相机拍照,将样品的色度、亮度等信息与对应浓度建立联系,应用于宽浓度范围测定和多组分体系分析。相比传统的分光光度法,深度学习辅助的比色法对单组分体系、多组分体系检测能力均有显著提升,对KMnO4单组分体系的浓度检测范围从1×10-5~9×10-4 mol/L拓宽到1×10-6~8×10-2 mol/L,对Co2+、Ni2+多组分体系的检测浓度范围也从1×10-2~1×10-1 mol/L拓宽到1×10-2~1.0 mol/L,并且检测效率显著提高,建立宽范围浓度样品的检测模型,可对后续未知样品浓度进行快速检测,为家庭临床检测与野外监测工作,提供了快速、便捷的定量分析手段。

关键词: 分光光度法, 视觉仿生, 角形比色皿, 深度学习, 宽浓度范围分析, 多组分体系

Abstract:

The application of spectrophotometry is limitted by its instrumental operation and serial concentration sample investigations for obtaining the concentration dependence linear diagram. With enormous and increasing requirements of daily health or clinic detection and field environment monitoring, rapid and facile approach with accessible equipment is desired. Inspired by human optesthesia chroma recognition, a deep learning assisted photographing colorimetry method was designed to achieve rapid colored multi-analytes quantitative analysis in this paper. The chroma, brightness and photographic information of the samples could be corresponded with their concentrations. Compared with the traditional spectrophotometry, the deep learning assisted photographing colorimetry improved the detection ability of single component system and multi-component system, which not only significantly widens the concentration detection range, but also improves the detection ability. The detection range of KMnO4 single component system was promoted from 1×10-5~9×10-4 mol/L of spectrophotometry to 1×10-6~8×10-2 mol/L of photographic colorimetry. The detection range of Co2+ and Ni2+ multi-component system was promoted from 1×10-2~1×10-1 mol/L of spectrophotometry to 1×10-2~1.0 mol/L of photographic colorimetry. Based on the big-data base of deep learning assisted photographing, the concentration of subsequent unknown samples can be quickly detected, which provides a fast and convenient quantitative analysis means for family clinical detection and field monitoring.

Key words: Spectrophotometry, Optesthesia inspired, Angular cuvette, Deep learning, Wide-range concentration analysis, Multi-analysis

中图分类号: