All Case Studies
Computer Vision
Satellite Image Matcher & Coordinate Projector
Align aerial & reference satellite images and project pixels to real coordinates.
2025
OVERVIEW
The Project
A desktop application that loads high-resolution aerial and reference satellite images, performs SIFT feature matching, computes a homography, and projects aerial pixel coordinates onto a geo-referenced map — a lightweight GIS-inspired tool for education and field applications.
Objectives
Tools & Technologies
PythonOpenCVSIFTNumPyPillowTkinter
METHODOLOGY
The Approach
1
Load a geo-referenced reference image (TIFF) and resized aerial images.
2
Compute SIFT features; match keypoints with Lowe's ratio test.
3
Estimate homography robustly with RANSAC.
4
On click, project the point to the reference image and convert pixel → lat/lon.
OUTCOME
Results & Learnings
Key Learnings
- Normalization, preprocessing, and good point selection drive matching quality.
- RANSAC is essential to reject outliers in homography estimation.
