Air pollution assessment method based on IGAN-CNN model

Abstract

The invention discloses an air pollution assessment method based on an IGAN-CNN model. The method comprises the following steps: step 1, acquiring a satellite image; 2, carrying out image enhancementprocessing; 3, carrying out data sample expansion: carrying out sample expansion on the satellite image in the step 2 by utilizing a generative adversarial network; 4, off-line modeltraining; and 5, on-line model application: applying the IGAN-CNN model obtained by training in the step 1 to 4 in an on-line manner so as to evaluate the air pollution degree of an image shot by a satellite in real time. The method provided by the invention has high evaluation precision and generalization performance, and has high practical application value.

Description

technical field [0001] The invention relates to the field of air pollution degree assessment, in particular to an air pollution assessment method based on an IGAN-CNN model. Background technique [0002] my country is the largest developing country in the world. In recent years, the economic and social development has been fruitful, the urbanization process has been steadily promoted, the degree of industrialization has been continuously improved, and people's living standards have been comprehensively improved and enhanced. However, the problem of air pollution left over from historical accumulation has become increasingly prominent and has become a key factor restricting China's sustainable development. In order to get rid of the plight of air pollution, the Chinese government has been actively using financial and administrative means to control air pollution. The "Environmental Protection Tax Law" has been officially implemented in 2018; Carry out research on urban air p...

AI Technical Summary

Problems solved by technology

[0004] Domestic Chinese invention patents involving the classification and identification of air pollution levels include "A Method for Air Pollution Estimation Based on Satellite Images" CN201910238150.9, announcement number CN109978862A, and announcement date 20190907. In this patent, threshold filtering is performed on infrared images, and then the The threshold-filtered image is subjected to key point detection, the overlap rate is calculated, and the obtained detection key points are deduplicated to obtain the scattered detection key points and detection range; For the detection key points whose pixel value is lower than the threshold, restore the remaining detection key points and detection range in the original image, but the determination of the filtering threshold in this patent may be different due to different images, that is, it is not adaptive

Images