Intelligent monitoring of water quality based on image analytics

Owing to the limitations of the spatial arrangement of monitoring stations and time acquisition of satellite remote sensing images, the water quality monitoring of rivers, especially small- and medium-sized rivers, cannot be satisfied in terms of time and space continuity. In this study, we propose a standardization method for the camera images derived at different locations on different time considering the influences of light, shadows, reflections, ripples, leaves and so on. After this algorithm is established, an estimation algorithm based on the comprehensive appearance of water body is proposed, which has the potential of realising real-time, mobile, and continuous monitoring of water quality with low costs. The test results showed that the accuracy of the model was quite high compared to the results of the hydrological monitoring stations. Compared with the single-point detection of water quality monitoring stations, this method has advantages in terms of dynamic detection and small- andmedium-sized water body detection, which can serve as a supplement to traditional detection. •A model for real-time estimation of water quality using water colour is constructed.•The image information dispersion degree is proposed to unify the colour and brightness of sample images.•The application of the model shows the continuous dynamic monitoring capability of water quality of the proposed model.•It complements the traditional monitoring and satellite remote sensing methods with higher temporal-spatial resolution.