That’s right folks– our first Food for Thought of the semester is right around the corner! Join us on February 21st from 12-1 PM for free lunch and an interesting talk about Geography, satellite imagery, and population estimates! Read the note from Professor Engstrom for more details, and sign up on the form below that!
People and Pixels: Mapping Population Characteristics using Satellite Imagery
This presentation will cover using satellite imagery to map variations in population
characteristics including poverty, income, and population density at multiple spatial scales
within a range of countries. My recent research has focused on extracting contextual features
from satellite imagery and determining their ability to improve estimates of population
characteristics in low and middle income countries at wide range of spatial scales. Contextual
features can be defined as the statistical quantification of edge patterns, pixel groups, gaps,
texture, and the raw spectral signatures calculated over groups of pixels or neighborhoods.
These features were originally designed for facial recognition software and have been adapted
to use with satellite imagery. Results have found that these features are highly correlated with
poverty, income, and population in Sri Lanka, Belize, and Ghana at scales from cities to entire
countries.
