Detecting the Level of Air Pollution: A Digital Image Processing Approach

There are a few particle monitoring programs implemented in the US, such as AirNow and the IMPROVE (Interagency Monitoring of Protected Visual Environments); however, these two are limited to certain locations and use professional laboratory equipment. IMPROVE only focuses on non-urban sites while AirNow excludes most of Africa, Russia, and Brazil, in addition to only updating once every two hours [9]. Just focusing on certain locations only provides a small part of the bigger picture because different pollutants are scattered across the entire globe, which also varies based on weather. Therefore it is important to frequently monitor as much land as possible in order to analyze the air quality of different locations across the globe. Whereas programs like IMPROVE and AirNow monitor air quality at specific locations and uses professional lab settings, this research method can measure air quality at any site using just a laptop and a digital camera. In my research, digital image processing is tested for its suitability in detecting the level of air pollution in a digital photograph.The spectral content of an image is analyzed and compared with the corresponding Air Quality Index (AQI) for that location. The goal is to reproduce what is normally costly and labor intensive chemical analysis with a less expensive and quicker digital spectral photoanalysis. This research is based on how particles such as carbon monoxide and nitrogen dioxide in air reflect, refract, and absorb the visible light spectrum. The sky is blue because the molecules in air scatter blue light more than they scatter red and green light. The particles in clean air are more effective at scattering higher frequency visible light (blue-colored light) while the lower frequency portion of the visible light spectrum (green and red) passes through a clean atmosphere with less refraction. However, when pollutants are present, the sky will appear less blue because pollutant particles are larger and tend to scatter longer wavelengths (green and red), thereby giving polluted sky a more yellow or brown color. Photographs from different locations are taken at the same time as when AQI is determined by local regulatory agencies and stored as arrays of RGB values for each pixel. These values are then graphed in a histogram of three channels, and the amount of overlap between the colors is calculated. Since pollutants in the atmosphere scatter red and green light more, it results in higher red and green pixel color values with lower blue pixel values, which leads to overlapping areas of the three color channels. A strong correlation between area of overlap and reported AQI demonstrates that the area of overlap can successfully be used measure air quality. The results of my research prove the positive linear correlation between the amount of overlapped pixels and the AQI, and such correlation is developed into an algorithm that generates the AQIs of new locations. The algorithm is very accurate, as demonstrated by the correlation coefficient, r=0.979.






Post a Comment

Be the first to comment on this article!

bRealTime banner ad on the left side
Site Feedback