Magellan Radar Demo - ML-Powered Surface Analysis | by Henrik Hargitai & Viktor Somogyi for HUN-REN / ESA
Processing...
Error
Click on the image to analyze a local patch
Click "Run Processing" to generate the image
Feature Space Visualization
Select Region of Interest
Drag to move the ROI box. Drag corners to resize.
Position:X: 0, Y: 0
Size:512 x 450 px
Minimum size: 128x128 pixels
Venus SAR Analysis Pipeline
Data flow: SAR data from NASA/USGS Magellan mission is preprocessed (normalization, despeckling),
then texture features are extracted using GLCM/Haralick methods. K-Means clustering groups
similar texture patches into surface units. PCA projects the feature space to 2D for visualization.
This is unsupervised exploration - no geological labels are assigned.
The tool proposes preliminary surface units based on texture similarity.