Research on pollutant identification methods in paper mill wastewater based on multisource spectral image fusion

Abstract

Intelligent algorithms are increasingly applied in identifying pollutants within complex environments, demonstrating particular advantages in fusing hyperspectral and visible light images. In order to solve the problems of multiple types of pollutants, severe spatial mixing, and overlapping spectral characteristics, a new approach to identification of pollutants based on multi-source spectrum image fusion is proposed. This method first uses Principal Component Analysis (PCA), Discrete Wavelet Transform (DWT), and CBAM-DenseFuse (DWT), and attention enhancement fusion strategy (CBAM-DenseFuse) to realize high resolution image fusion. Then, it combines a spatio-spectral joint feature design with the Transformer and ResNet fusion architecture to extract the outstanding features from the mixed regions. In order to solve this problem, SpecAugment, MAML meta-learning, and Mixup are introduced to improve the generalization performance of the model. The results show that the proposed model is superior to the existing methods in terms of spectral angle error and edge retention, and achieves 92.0% classification accuracy. The study provides an efficient way to accurately identify the pollutants in the waste water of the paper mill, which has a promising future in the field of engineering application.