Haze Grading Using the Convolutional Neural Networks

As an air pollution phenomenon, haze has become one of the focuses social discussion. Research into causes and concentration prediction is significant, forming basis prevention. The inversion Aerosol Optical Depth (AOD) based on remote sensing satellite imagery can provide a reference for major pollutants in haze, such as PM2.5 PM10 concentration. This paper used to study problems chose PM2.5, primary pollutants, research object. First, we conventional methods perform AOD images, verifying correlation between PM2.5. Subsequently, simplify parameter complexity traditional method, proposed using convolutional neural network instead method constructing level model. Compared with aerosol depth inversion, found that networks higher through more simplified image processing process. Thus, it offers possibility researching managing networks.