Estimation of PM10 Concentration with Features Extraction by Digital Image Processing

Since the city’s air quality has declined, developing an efficient and convenient method to measure PM10 concentration has become an interesting research topic. There is a correlation between the characteristics of air pollution and the reducing in visibility which can be estimated by image processing methods. To estimate the concentration of PM10 by image processing is the main purpose of this paper. The automatically selecting RoI is applied for extracting the features from the image. The high frequency information, portion of the non-scattered light, relative humidity and PM10 concentration are regressed by Support Vector Regression. The real images and two indices are used to show the performance of the estimation. According to the experimental results, R2 is obtained greater than 0.71 and the RMSE is less than 16.1.