RoadwayRegularizer

class rapidtools.processing.RoadwayRegularizer(min_width_ft: float = ..., min_network_length_ft: float = ..., output: Literal['both'] = ...)

class rapidtools.processing.RoadwayRegularizer(min_width_ft: float = ..., min_network_length_ft: float = ..., *, output: Literal['polygons', 'centerlines'])

class rapidtools.processing.RoadwayRegularizer(min_width_ft: float = ..., min_network_length_ft: float = ..., output: str = ...)

Bases: Generic[_OutT]

Pipeline component to regularize jagged raster-to-vector road polygons.

This regularizer processes raw input geometries, extracts a Voronoi-based centerline skeleton, heals the graph, merges collinear segments, and statistically samples widths. It outputs in-memory PhysicalAssetCollection objects for both the centerlines and the reconstructed polygons.

Parameters:

• min_width_ft (float, optional) – The minimum enforced width for any regularized road segment in feet. Defaults to 22.0.

• min_network_length_ft (float, optional) – The minimum cumulative length a connected skeleton component must possess to be considered a valid road network. Any disconnected stub networks shorter than this value will be filtered out. Defaults to 100.0.

Example

>>> from rapidtools.core import PhysicalAssetCollection
>>> from rapidtools.processing import RoadwayRegularizer
>>>
>>> raw = PhysicalAssetCollection.from_geojson('roads_raw.geojson')
>>> regularizer = RoadwayRegularizer(
...     min_width_ft=22.0, min_network_length_ft=150.0
... )
>>> centerlines, polygons = regularizer(raw)
>>> centerlines[0].attributes
{'asset_type': 'road_centerline', 'width_ft': 24.5, 'azimuth_deg': 91.2}

Methods

process

Execute the regularization logic and return asset collections.

process(input_assets: PhysicalAssetCollection) → tuple[PhysicalAssetCollection, PhysicalAssetCollection]

Execute the regularization logic and return asset collections.

The steps are: project to a local feet-based CRS, drop polygons smaller than MIN_POLYGON_AREA_SQFT, fuse near-touching polygons, skeletonize with a Voronoi diagram, heal/prune the skeleton graph, remove short stub networks, merge collinear segments, sample widths and azimuths, and finally rebuild WGS84 centerline and polygon assets.

Parameters:

input_assets (PhysicalAssetCollection) – The raw input road polygons (WGS84).

Returns:

tuple[PhysicalAssetCollection, PhysicalAssetCollection] – A tuple containing:

• centerlines_collection: Assets representing the road centerlines (LineString geometries) with the attributes 'asset_type': 'road_centerline', 'width_ft' and 'azimuth_deg'.

• polygons_collection: Assets representing the regularized polygons with 'asset_type': 'road_polygon' and the same width/azimuth attributes.

Both collections are empty when the input is empty or when no polygon exceeds the minimum area.

Example

>>> centerlines, polygons = RoadwayRegularizer().process(raw_roads)
>>> polygons[0].attributes['asset_type']
'road_polygon'