Related Experiment Video
Updated: Sep 13, 2025
08:13
A Filter-based Surface Enhanced Raman Spectroscopic Assay for Rapid Detection of Chemical Contaminants
Published on: February 19, 2016
9.5K
Detection of surface water pollution with full absorption spectroscopy: Integrating 2D image conversion and ResNet
Chi Yao1, Hongyang Wang1, Siying Liu1
1School of Computer Science and Engineering, University of Electronic Science and Technology of China, Chengdu, 611731, China.
Summary
A new MGR model accurately classifies surface water pollution using spectral data. This advanced method improves water quality monitoring and pollution source identification for sustainable resource management.
Area of Science:
- Environmental Science
- Analytical Chemistry
- Data Science
Background:
- Industrialization and urbanization degrade water environments, necessitating advanced monitoring.
- Traditional water quality methods are insufficient for complex pollution challenges.
- Surface water pollution poses risks to health, ecosystems, and economy.
Purpose of the Study:
- To develop and validate a novel MGR (MSC-GAF-ResNet) model for surface water pollution classification.
- To enhance the efficiency and accuracy of water quality monitoring.
- To enable effective pollution source tracing for better environmental management.
Main Methods:
- Utilized full-spectrum absorption data for surface water samples.
- Applied Savitzky-Golay filtering and Multiplicative Scatter Correction for data preprocessing.
- Employed Gramian Angular Field (GAF) for spectral data conversion and a Residual Convolutional Neural Network (ResNet) for classification.
Main Results:
- The MGR model achieved 97.7% classification accuracy in identifying surface water pollution.
- Demonstrated high computational efficiency compared to traditional machine learning methods (SVM, LSTM).
- Integration of metadata facilitated effective pollution source tracing.
Conclusions:
- The MGR model offers an efficient and accurate approach to surface water quality monitoring and pollution classification.
- This method supports sustainable water resource management through improved monitoring and emergency response.
- The study highlights the potential of spectral data combined with deep learning for environmental applications.
