Vectorize scanned topographic maps using interactive HSV colour-range classification
Converts scanned raster maps (JPEG/PNG/GeoTIFF) into polygon vector layers.
Workflow:
1. Draw a rectangle on the map to sample colours
2. K-means auto-clustering splits the region into colour groups
3. Adjust HSV ranges and tolerance per class
4. Run: median pre-filter → HSV classification → morphological clean-up → GDAL Polygonize
5. Result: a polygon vector layer with class_id and class_name attributes
Key features:
- HSV colour space matching (perceptually uniform, easy to tune)
- K-means auto-classification (2–25 clusters)
- Median filter preprocessing (removes JPEG noise without blurring edges)
- Morphological opening/closing (removes specks, fills gaps)
- Handles non-standard GeoTransform (rotated/flipped rasters via GDAL Warp)
- Live preview overlay while editing HSV ranges
- Export/import colour classes as JSON
- SciPy optional (NumPy fallback for all operations)
Developed as part of a research project under the supervision of
Pavel P. Lepekhin (Лепехин Павел Павлович).
Plugin Tags