Optical Flow for Motion Analysis
Estimating motion from sequential images using dense optical flow in OpenCV and applying the same ideas to satellite data.
Example video used for motion analysis.
Optical flow estimates apparent motion between sequential frames by assigning a displacement vector to image locations. It is a foundational idea in computer vision because it converts visual change into a measurable field of direction and magnitude that can be analyzed quantitatively.
For this project, I used dense optical flow in OpenCV, based on Gunnar Farneback's method, to compute motion across entire frames rather than only at a sparse set of keypoints. I visualized the direction and magnitude of motion using HSV color encoding and vector-field outputs to make the results easier to interpret.
I also applied the same concept to sequential sea-surface temperature satellite imagery. In that context, apparent temperature motion can act as a proxy for surface movement, allowing optical-flow methods to support exploratory analysis of ocean dynamics from remotely sensed data.
