developing a particulate matter (pm)-indicating …digital+assets/c...1. introduction 2. proposal :...
TRANSCRIPT
1. Introduction
2. Proposal : Smart Phone + Image Analysis
4. App Development:Particular Matter Detector
Developing a Particulate Matter (PM)-indicating Smartphone AppMengxuan(Billy) Cai a, Gang (Ian) Chen b, Arthur W.H. Chan b
a Department of Applied Science & Engineering, b Department of Chemical Engineering & Applied Chemistry, University of Toronto, Toronto, ON, Canada
6. Conclusion
CENTRE FOR GLOBAL CHANGE SCIENCE
References and Acknowledgements
Widespread• Around 3 billion people cook
and heat their home using an
open fire or simply burning
biomass and coal1
• PM concentration: 500-1000
µg/m3 during cooking
Health Impacts• Over 4 million people die
prematurely from illness
attributable to the household air
pollution from cooking with
solid fuels1
• Collect the samples (N>3000) by SHARP and scan it by scanner
• Extract RGB values by photo processing
• Establish the correlation between the darkness of the filter paper and
the mass of PM deposited on it
• Plug in the algorithm developed into the app.
References:
1.Kirk R. Smith et al.(2017, July). Millions Dead: How Do We Know and What Does It Mean? Methods Used in
the Comparative Risk Assessment of Household Air Pollution.[Online].Available:
http://www.annualreviews.org/doi/pdf/10.1146/annurev-publhealth-032013-182356
2.World Health Organization.(2017, July). Indoor air pollution and household energy[Online].Available:
http://www.who.int/heli/risks/indoorair/indoorair/en/
Acknowledgements
The authors acknowledge funding through the Center For Global Change Science (CGCS)
3. Method
Air Flow 1 m3/hr
Figure 3. Schematic Diagram of
Synchronized Hybrid Ambient Real-
time Particulate (SHARP) Monitor
Develop an affordable and relatively accurate PM sensor
Figure 1. Health Impact of Solid Fuel
Exposures1
Current Challenges in
PM-sensor Market
Avg. Price = $100 USD = 2.1
months’ household income
Figure 4. CV Filter
Paper Color Picker
Figure 5. MATLAB Machine
Learning Toolbox
Figure 7. The Application Work Flow
Platform: XCode 8
Language: Swift 3.1
Image Analysis Library: OpenCV
Other Open Source Library: Cocoa pod
Image Library
Custom Slider:• Perform edge detection
Figure 6. App Icon
Figure 8. The First View Figure 9. User Guide
Figure 10. The Second View Figure 11. Final View
5. Future Work
PM Exposure Detection for Face Masks’ Sampling
Figure 17. Schematic Diagram of Breathing
Machine
Figure 18. Ambient Particles Concentrator
at Gage Occupational Health
Image Analysis:
• Try other white balance algorithms
• Extract other relevant parameters (e.g., lightness, and saturation)
• iOS
• Less variability on camera
OpenCV:
• White Balance Algorithm: Gray World
Assumption
𝑅average = 𝐺 average = 𝐵 average
• Edge Detection
Custom Slider:• Disk Colour: Using AQI Colour
Estimated PM Exposure: g
0-50 50-100 100-150 150-200 200-300 300-500
Good Moderate Unhealthy
for
Sensitive
Group
Unhealthy Very
Unhealthy
Hazardous
The Second ViewCustom Camera: Paparazzo
• Take photos with the highest
quality
• Back & Front camera
• Flash light
• Rotate the image
• Select canvas & crop image
Figure 12. Photo Capture on Paparazzo.
(a) Camera; (b) Image Cropping
Figure 13.
Origin image
Figure 14.
Grayscale image
Figure 15.
Monochrome image
Figure 16. Air Quality Index (North America)
Filter Paper
• Particulate Matter Samples
Raw Image
• Custom Camera
Image Process
•Auto-White Balance Adjustment
Extract Colour Information
• Edge Detection
Estimate PM Exposure
•Model Build-up
Health Impact Info
•Air Quality Index
• Smartphone image analysis offers a cheap and relatively accurate PM
indicating method
• Future commercialization will not only save millions of peoples’ lives, but
also provide enough evidences for policy making.
• Developing countries
• Free smart phone app
• Easily distributed
Figure 2. Deaths From Indoor Smoke From Solid Fuels2
The Final View
a
b Air Quality Index (North America Standard):