Identification for water quality based on color characteristics

• Published in: IOP conference series. Earth and environmental science Vol. 983; no. 1; pp. 12075 - 12083
• Main Authors: Wang, Jiangang, Zhai, Zhengang, Zhu, Yunya, Zhang, Li, Fang, Xusheng
• Format: Journal Article
• Language: English
• Published: Bristol IOP Publishing 01.02.2022
• Subjects: Color; Digital imaging; Historical account; Image acquisition; Laboratory methods; Learning algorithms; Machine learning; Support vector machines; Water quality; Water resources
• ISSN: 1755-1307; 1755-1315
• DOI: 10.1088/1755-1315/983/1/012075

It’s great significance for protection of water ecological and water resources to identify water quality rapidly and conveniently. In the past time, water quality was test and monitored with traditional laboratory methods, which was hard to meet the requirements of urgent demand. A rapid and convenient method for the identification of water quality based on machine learning was used in this study. By sampling and photographing, the image of water was acquired. Then nine dimensional digital information features of the color information were obtained by the moment method. Based on the historical data and expert experience, a support vector machine (SVM) model was successfully built and well trained. Then the model was verified with the test data, and the accuracy reaches 95%, which proves this method has good effect and high precision. This work will generate fresh insight into water quality identification and contribute to water resources protection.