Abstract:Continuous air pollution monitoring is fundamental for exposure assessment and management. Traditionally, air quality is measured by standard air quality monitoring stations characterized by accuracy and reliability. However, these stations are expensive and bulky to setup and maintain, thus sparsely distributed. Our research aims to establish a new low-cost and portable method for estimating particulate matter levels, using digital cameras and imagery response to light scattering by particles. For that purpose, more than 18,000 daylight images of Haifa Bay scene were acquired from August 2012 to March 2013, imaged from Technion, Israel. Since image metrics, including color content and quality measures, yielded low correlations to simultaneously measured PM2.5 levels (R2 ≤ 0.4) we further considered influences of illumination and meteorology conditions on image metrics. To this end, the effect of possible modifying variables, such as the solar illumination angle and the ambient relative humidity, were studied. After stratifying the dataset by different modifiers, two of the image metrics produced promising results: Perceptual Blur Metric, and Cyan content in the image. Multiple linear regression models over the stratified data with those image metrics, yielded much higher determination coefficients. As an example, models over data stratified by illumination angles to 10 quantiles produced 0.445 ≤ R2adj ≤ 0.656. Additional stratifying of the dataset by the different months improved the models adjusting up to R2adj = 0.967 in certain strata. These suggests a great potential for the proposed method as a way to estimate PM2.5 concentrations at ground level. A comprehensive analysis is ongoing, including additional datasets of more than 3,000 images from Brisbane, Australia. Using the entire data, we hope to develop a solid model for estimating PM levels based on image quality metrics.

Real-time air quality estimation based on color image processing

Effective and Efficient Photo-Based PM2.5 Concentration Estimation

PM2.5 Concentration Estimation Based on Image Quality Assessment

Vision-Based PM$_{2.5}$ Concentration Estimation with Natural Scene Statistical Analysis

PM2.5 Estimation Based on Image Analysis.

Image-Based PM2.5 Estimation and Its Application on Depth Estimation

Estimate Air Quality Based On Mobile Crowd Sensing And Big Data

Particle Pollution Estimation from Images Using Convolutional Neural Network and Weather Features

Highly Efficient Picture-Based Prediction of PM2.5 Concentration

Real-time image-based air quality estimation by deep learning neural networks

A PM2.5 concentration estimation method based on multi-feature combination of image patches

Air Quality Evaluation Based on Local Normalized Image Contrast

Extremely Efficient PM2.5 Estimator Based on Analysis of Saliency and Statistics

Hybrid Measurement Of Air Quality As A Mobile Service: An Image Based Approach

Crowdsourced Image Driven PM2.5 Estimation Based on Hybrid 3-Channel Feature Map

Air Quality Controlling-Oriented Highly Efficient Method For Monitoring Particulate Matters

Estimation of Atmospheric PM2.5 Based on Photos and Deep Learning.

Assessing Exposure to Ambient Fine Particulate Matter Based on Digital Imagery

Particle Pollution Estimation Based on Image Analysis.

Real-Time Air Pollutants Rendering Based on Image Processing

Method for estimating PM2.5 value of haze based on image