Abstract: With the gradual development of the rural economy, the industrial sewage discharged from the factories in the surrounding countryside and the sewage generated in the process of rural production and life have caused great pollution to the rural water environment. In order to help improve the water environment, monitor changes in the water environment, and build beautiful villages, this paper classified the water quality in rural areas based on image processing technology and machine learning algorithms. By cutting the collected water quality images and extracting the color moment features of the images, the water quality classification model was established as a sample, and the water quality and water color classification were performed using decision trees and support vector machines respectively. The classification results was used to judge whether the water quality was polluted, and timely measures were taken to provide support for the treatment of rural water pollution. The classification results showed that the model accuracy of support vector machine was 96%, the model accuracy of decision tree was 83%, and the recognition accuracy of support vector machine was higher than that of decision tree.

Key words: smart agriculture, water quality classification, prediction and warning, rural water resources, water pollution