COMPUTER VISION-BASED ALGORITHM FOR DETECTING FLOATING WASTE AND ESTIMATING ITS AREA AT PUMPING STATIONS

Abstract

The efficient and stable operation of pumping stations largely depends on the hydraulic conditions at the water inlet. During practical operation, various wastes accumulating on the water surface can negatively affect the operation of pumping units and impede the free movement of water flow. As a result, hydraulic resistance increases, additional load is created on the pumping system, and its efficiency decreases. In this research work, an approach based on video monitoring and image processing technologies was developed to detect wastes accumulating on the water surface at pumping stations. In the proposed method, wastes on the water surface are separated using image segmentation and object recognition algorithms and their area in pixels is calculated. A mathematical model was proposed that allows estimating the share of the area occupied by wastes based on the obtained image data. This approach serves to determine the amount of wastes on the water surface in real time, automate the process of their monitoring, and ensure the efficient operation of pumping stations.