A MATLAB application for feature extraction, enhancement, spectral analysis, and segmentation of 2D geoscience data — satellite/aerial imagery, geophysical grids, and point datasets.
Author: Bahman Abbassi, PhD · University of Québec in Abitibi-Témiscamingue (UQAT) · Institut de Recherche en Mines et Environnement (IRME)
Important
This repository distributes the licensed, compiled (P-code) version of SFES2D. The source code is not included. Running the software requires a license file tied to your computer — see Licensing.
SFES2D is a complete workbench for extracting, enhancing, and analyzing features in 2D geoscience data — multiband imagery, geophysical grids (magnetic, gravity, radiometric, DEM, …), and scattered X,Y,Z point datasets. Its capabilities span the full workflow across six tabs:
| Stage | Tab | Capabilities |
|---|---|---|
| Import & georeference | Data Sets | Load multiband/RGB imagery, single grids, and scattered points; interpolate onto a common grid; pre-condition with isotropic, anisotropic, structure-guided, bilateral, multitap, or guided filters. |
| Multivariate feature extraction | Spatial | Decorrelate and compress correlated bands/layers with PCA and ICA (kurtosis- and negentropy-based). |
| Multiscale directional spectral analysis | Spectral | 2D Continuous Wavelet Transform with several mother wavelets (Mexican hat, derivatives of Gaussian/Poisson, complex Gaussian/Poisson); automated saliency/dissimilarity feature selection; analytic-signal masking; potential-field upward continuation. |
| Representation & segmentation | Represent | Build and optimize RGB composites; segment with K-means or Gaussian Mixture Models in RGB or LAB color space, with automatic cluster-number selection. |
| Lineament extraction | Lineaments | Map faults/fractures/contacts with two engines — Topologic (directional step-filter / GRM) and Cubic (Hessian ridge + optional deep-learning denoising) — producing polylines, density maps, and rose-diagram statistics. |
| Export | Export | Save any layer or result as CSV, GeoTIFF, or ESRI Shapefile for use in a GIS. |
Together these make SFES2D useful well beyond structural mapping — for image enhancement, data fusion, spectral decomposition, and unsupervised classification of geoscience datasets.
| Component | Minimum | Recommended |
|---|---|---|
| MATLAB | R2019b | R2021a or later |
| RAM | 8 GB | 16 GB or more |
| CPU | Dual-core | Quad-core or better |
Toolboxes
- Image Processing Toolbox — required
- Statistics and Machine Learning Toolbox, Mapping Toolbox — recommended
- Deep Learning Toolbox, Parallel Computing Toolbox — optional features
- Download or clone this repository to a folder on your computer.
- Launch MATLAB and add the folder to the path:
addpath(genpath('path\to\SFES2D')); - Start the app:
SFES2D
- Launch
SFES2D. (First time only: if a "License Required" window appears, install your license — see Licensing.) - Data Sets — load an image or grid, enter Lon/Lat bounds, and click Polygonize.
- (Optional) Spatial / Spectral — enhance with PCA/ICA and the 2D CWT.
- Lineaments — tick your inputs, click Merge, then run an extraction engine.
- Export — save results to CSV, GeoTIFF, or Shapefile.
See the user manual for the full tutorial.
SFES2D is licensed software, locked to a single computer via its network-card (MAC) address, with an expiry date. On first launch, if no valid license is found, a License Required window shows your computer's MAC address and instructions.
To request a license, send the MAC address shown (or click License in the app menu, then Copy MAC to Clipboard) to:
- 🎓 Free academic / research use:
bahman.abbassi@uqat.ca— include a short description of your research. Publications using SFES2D must cite the associated papers. - 🏢 Commercial / industry use: Prof. Li-Zhen Cheng,
li_zhen.cheng@uqat.ca— terms set by UQAT.
Save the received SFES2D_license.lic file into the SFES2D application folder and restart the app.
See SFES2D_User_Manual.md for the full user guide — tab-by-tab reference, parameter tuning, worked example, and mathematical background.
Feature extraction and analysis were performed using SFES2D (Spectral Feature Extraction and Selection 2D) v6.1, a MATLAB GUI for multivariate decomposition (PCA/ICA), 2D continuous wavelet analysis, image segmentation, and lineament mapping of geoscience data. B. Abbassi, UQAT.
Proprietary. See LICENSE. © University of Québec in Abitibi-Témiscamingue (UQAT). All rights reserved.
Bahman Abbassi, PhD — bahman.abbassi@uqat.ca