Laguerre-Gauss Preprocessing: Line Profiles as Images Features

Code implementation of the results presented on Laguerre-Gauss Preprocessing: Line Profiles as Images Features for Aerial Images Classification.

https://arxiv.org/abs/1912.06729

Usage

Using Python (v3.7.3) run the files MLP.py, kNN.py and CNN.py to test the results on the trained models using a subsample of the test set of aerial images.

Subsample of test set

The folder @sample_test_data@ contains a randomly chosen subsample of the test set for aerial images classification. The images are 64x64 RGB classified as 0 (nothing of interest: populations, rivers, forest, etc.) or 1 (something of interest: heavy machinery, boats, deforestation, etc.).