Intelligent monitoring method for agricultural nonpoint source pollution based on remote sensing image changes

To address issues related to insufficient monitoring coverage, inaccurate measurement of pollution concentration, and inadequate load monitoring using traditional methods, an intelligent monitoring method for agricultural non-point source pollution based on remote sensing image changes is proposed. Utilising Landsat ETM+ remote sensing imagery, a multi-layer perceptron algorithm is employed to detect and identify agricultural non-point source pollution areas. The Min-Cut algorithm is used for remote sensing image segmentation of agricultural non-point source pollution targets. Remote sensing image changes are detected based on the segmentation results and difference diagram. Intelligent monitoring of agricultural non-point source pollution is achieved by combining the detection results of remote sensing image changes with relevant monitoring indicators. Experimental findings demonstrate that the proposed method achieves a monitoring coverage rate exceeding 96.3%, the mean accuracy of pollution concentration monitoring is 97.1%, the maximum accuracy of pollution load monitoring is 99.3%.