Apply some light pre-processing of a geodataframe.
Namely, remove invalid polygons, empty geometries and add bounds to the gdf.
Parameters:
| Name |
Type |
Description |
Default |
gdf |
GeoDataFrame
|
|
required
|
min_size |
int
|
The minimum size of the polygons in pixels.
|
None
|
Returns:
| Type |
Description |
Union[GeoDataFrame, None]
|
gpd.GeoDataFrame:
The pre-processed gdf or None if input gdf is empty or None.
|
Examples:
>>> from cellseg_gsontools import pre_proc_gdf
>>> gdf = pre_proc_gdf(gdf, min_size=100)
Source code in cellseg_gsontools/utils.py
| def pre_proc_gdf(
gdf: gpd.GeoDataFrame, min_size: int = None
) -> Union[gpd.GeoDataFrame, None]:
"""Apply some light pre-processing of a geodataframe.
Namely, remove invalid polygons, empty geometries and add bounds to the gdf.
Parameters:
gdf (gpd.GeoDataFrame):
Input geodataframe.
min_size (int, optional):
The minimum size of the polygons in pixels.
Returns:
gpd.GeoDataFrame:
The pre-processed gdf or None if input gdf is empty or None.
Examples:
>>> from cellseg_gsontools import pre_proc_gdf
>>> gdf = pre_proc_gdf(gdf, min_size=100)
"""
if gdf.empty or gdf is None:
return gdf
# drop invalid geometries if there are any after buffer
gdf.geometry = gdf.geometry.buffer(0)
gdf = gdf[gdf.is_valid]
# drop empty geometries
gdf = gdf[~gdf.is_empty]
# if there are multipolygon geometries, explode them
if "MultiPolygon" in list(gdf["geometry"].geom_type):
gdf = gdf.explode(index_parts=False).reset_index(drop=True)
# drop geometries that are less than min_size pixels
if min_size is not None:
gdf = gdf[gdf.area > min_size]
# drop geometries that are not polygons
gdf = gdf[gdf.geom_type == "Polygon"]
try:
# add bounding box coords of the polygons to the gdfs
# and correct for the max coords
gdf["xmin"] = gdf.bounds["minx"].astype(int)
gdf["ymin"] = gdf.bounds["miny"].astype(int)
gdf["ymax"] = gdf.bounds["maxy"].astype(int) + 1
gdf["xmax"] = gdf.bounds["maxx"].astype(int) + 1
except Exception:
warnings.warn("Could not create bounds cols to gdf.", RuntimeWarning)
return gdf
|