Remote Sensing Image Processing

A hands-on, module-by-module portfolio of image-processing techniques applied to Sentinel-2 satellite imagery — built to demonstrate practical geospatial and remote-sensing skills in Python, from raster fundamentals up to full analysis workflows.

Each module is a self-contained folder of small, well-documented scripts. Shared logic lives in a reusable utils/ package, so the codebase stays clean as it grows. Built with rasterio, numpy, and matplotlib.

What this project demonstrates

• Geospatial data handling — reading and writing georeferenced rasters (GeoTIFF), preserving CRS and affine transforms, and interpreting metadata.
• Remote-sensing fundamentals — multi-band imagery, spectral bands, true-colour composites, and spectral indices such as NDVI.
• Image processing — contrast enhancement, filtering, edge detection, morphology, and segmentation (progressively added across modules).
• Clean software practices — modular structure, a shared utilities package to avoid code duplication, per-module documentation, and reproducible outputs.

Progress

#	Module	Status
01	Image Basics	✅ Done
02	Image Enhancement	🚧 Planned
03	Image Filtering	🚧 Planned
04	Edge Detection	🚧 Planned
05	Morphological Operations	🚧 Planned
06	Image Segmentation	🚧 Planned
07	Feature Extraction	🚧 Planned
08	Remote Sensing Applications	🚧 Planned

Repository structure

remote-sensing-image-processing/
├── data/
│   └── sample/sentinel2.tif      # sample Sentinel-2 scene (multi-band GeoTIFF)
├── utils/
│   ├── __init__.py
│   └── raster_io.py              # shared helpers: read, stretch, save PNG/GeoTIFF
├── modules/
│   ├── 01_image_basics/          # each module is a folder of scripts + a README
│   ├── 02_image_enhancement/
│   ├── 03_image_filtering/
│   ├── 04_edge_detection/
│   ├── 05_morphological_operations/
│   ├── 06_image_segmentation/
│   ├── 07_feature_extraction/
│   └── 08_remote_sensing_applications/
├── outputs/                      # generated images/rasters (git-ignored)
├── requirements.txt
├── .gitignore
└── README.md

Setup

python -m venv .venv
source .venv/bin/activate        # Windows: .venv\Scripts\activate
pip install -r requirements.txt

rasterio bundles its own GDAL via pip wheels, so no separate GDAL install is needed on most systems.

Usage

Run any script directly; each takes an optional image path and otherwise falls back to the sample scene. Generated files are written to outputs/.

python modules/01_image_basics/03_display_rgb_image.py
python modules/01_image_basics/06_save_processed_image.py path/to/your.tif

How it's built

The repo is organised so that adding a new technique is always the same small, predictable step:

• Modular layout. Every topic lives in its own numbered folder under modules/, each with a short README explaining the concept and listing its scripts. Numbering keeps the learning path in order and easy to browse.
• Shared utilities. Reusable logic — opening rasters, percentile contrast stretching, and saving PNGs and georeferenced GeoTIFFs — lives once in utils/raster_io.py. Scripts import from it (from utils import ...) instead of repeating boilerplate, which keeps each script short and focused on the technique it demonstrates.
• Reproducible outputs. Scripts read from a sample scene by default and write results to outputs/, which is git-ignored so generated files never clutter the repository.
• Consistent script pattern. Every script accepts an optional image path, documents its assumptions (e.g. band ordering), and can be run standalone.

To add a module: create the next numbered folder, drop in scripts that import from utils, write a short module README, and flip its row in the progress table above.

A note on band ordering

Sentinel-2 stacks don't have a universal band order. Scripts that need specific bands declare them as constants at the top (e.g. RED_BAND, NIR_BAND); adjust these to match your file. Run modules/01_image_basics/02_image_metadata.py first to confirm your band count and order.

Tech stack

Python · rasterio · NumPy · matplotlib

License

MIT