import copy
from functools import lru_cache
import numpy as np
from astropy import units as u
from astropy.coordinates import Angle, SkyCoord
from astropy.io import fits
from astropy.table import QTable, Table
from astropy.utils import lazyproperty
from astropy.wcs.utils import (
proj_plane_pixel_area,
proj_plane_pixel_scales,
wcs_to_celestial_frame,
)
from regions import (
CompoundSkyRegion,
PixCoord,
PointSkyRegion,
RectanglePixelRegion,
Regions,
SkyRegion,
)
import matplotlib.pyplot as plt
from gammapy.utils.regions import (
compound_region_center,
compound_region_to_regions,
regions_to_compound_region,
)
from ..axes import MapAxes
from ..core import Map, MapCoord
from ..geom import Geom, pix_tuple_to_idx
from ..utils import _check_width
from ..wcs import WcsGeom
__all__ = ["RegionGeom"]
[docs]class RegionGeom(Geom):
"""Map geometry representing a region on the sky.
The spatial component of the geometry is made up of a
single pixel with an arbitrary shape and size. It can
also have any number of non-spatial dimensions. This class
represents the geometry for the `RegionNDMap` maps.
Parameters
----------
region : `~regions.SkyRegion`
Region object.
axes : list of `MapAxis`
Non-spatial data axes.
wcs : `~astropy.wcs.WCS`
Optional wcs object to project the region if needed.
binsz_wcs : `float`
Angular bin size of the underlying `~WcsGeom` used to evaluate
quantities in the region. Default size is 0.01 deg. This default
value is adequate for the majority of use cases. If a wcs object
is provided, the input of binsz_wcs is overridden.
"""
is_regular = True
is_allsky = False
is_hpx = False
is_region = True
_slice_spatial_axes = slice(0, 2)
_slice_non_spatial_axes = slice(2, None)
projection = "TAN"
def __init__(self, region, axes=None, wcs=None, binsz_wcs="0.1 deg"):
self._region = region
self._axes = MapAxes.from_default(axes, n_spatial_axes=2)
self._binsz_wcs = binsz_wcs
if wcs is None and region is not None:
if isinstance(region, CompoundSkyRegion):
self._center = compound_region_center(region)
else:
self._center = region.center
wcs = WcsGeom.create(
binsz=binsz_wcs,
skydir=self._center,
proj=self.projection,
frame=self._center.frame.name,
).wcs
self._wcs = wcs
self.ndim = len(self.data_shape)
# define cached methods
self.get_wcs_coord_and_weights = lru_cache()(self.get_wcs_coord_and_weights)
def __setstate__(self, state):
for key, value in state.items():
if key in ["get_wcs_coord_and_weights"]:
state[key] = lru_cache()(value)
self.__dict__ = state
@property
def frame(self):
"""Coordinate system, either Galactic ("galactic") or Equatorial
("icrs")."""
if self.region is None:
return "icrs"
try:
return self.region.center.frame.name
except AttributeError:
return wcs_to_celestial_frame(self.wcs).name
@property
def binsz_wcs(self):
"""Angular bin size of the underlying `~WcsGeom`
Returns
-------
binsz_wcs: `~astropy.coordinates.Angle`
"""
return Angle(proj_plane_pixel_scales(self.wcs), unit="deg")
@lazyproperty
def _rectangle_bbox(self):
if self.region is None:
raise ValueError("Region definition required.")
regions = compound_region_to_regions(self.region)
regions_pix = [_.to_pixel(self.wcs) for _ in regions]
bbox = regions_pix[0].bounding_box
for region_pix in regions_pix[1:]:
bbox = bbox.union(region_pix.bounding_box)
try:
rectangle_pix = bbox.to_region()
except ValueError:
rectangle_pix = RectanglePixelRegion(
center=PixCoord(*bbox.center[::-1]), width=1, height=1
)
return rectangle_pix.to_sky(self.wcs)
@property
def width(self):
"""Width of bounding box of the region.
Returns
-------
width : `~astropy.units.Quantity`
Dimensions of the region in both spatial dimensions.
Units: ``deg``
"""
rectangle = self._rectangle_bbox
return u.Quantity([rectangle.width.to("deg"), rectangle.height.to("deg")])
@property
def region(self):
"""`~regions.SkyRegion` object that defines the spatial component
of the region geometry"""
return self._region
@property
def is_all_point_sky_regions(self):
"""Whether regions are all point regions"""
regions = compound_region_to_regions(self.region)
return np.all([isinstance(_, PointSkyRegion) for _ in regions])
@property
def axes(self):
"""List of non-spatial axes."""
return self._axes
@property
def axes_names(self):
"""All axes names"""
return ["lon", "lat"] + self.axes.names
@property
def wcs(self):
"""WCS projection object."""
return self._wcs
@property
def center_coord(self):
"""(`astropy.coordinates.SkyCoord`)"""
return self.pix_to_coord(self.center_pix)
@property
def center_pix(self):
"""Pixel values corresponding to the center of the region"""
return tuple((np.array(self.data_shape) - 1.0) / 2)[::-1]
@lazyproperty
def center_skydir(self):
"""Sky coordinate of the center of the region"""
if self.region is None:
return SkyCoord(np.nan * u.deg, np.nan * u.deg)
return self._rectangle_bbox.center
@property
def npix(self):
"""Number of spatial pixels"""
return (1, 1)
[docs] def contains(self, coords):
"""Check if a given map coordinate is contained in the region.
Requires the `.region` attribute to be set.
Parameters
----------
coords : tuple, dict, `MapCoord` or `~astropy.coordinates.SkyCoord`
Object containing coordinate arrays we wish to check for inclusion
in the region.
Returns
-------
mask : `~numpy.ndarray`
Boolean Numpy array with the same shape as the input that indicates
which coordinates are inside the region.
"""
if self.region is None:
raise ValueError("Region definition required.")
coords = MapCoord.create(coords, frame=self.frame, axis_names=self.axes.names)
return self.region.contains(coords.skycoord, self.wcs)
[docs] def contains_wcs_pix(self, pix):
"""Check if a given wcs pixel coordinate is contained in the region.
Parameters
----------
pix : tuple
Tuple of pixel coordinates.
Returns
-------
containment : `~numpy.ndarray`
Bool array.
"""
region_pix = self.region.to_pixel(self.wcs)
return region_pix.contains(PixCoord(pix[0], pix[1]))
[docs] def separation(self, position):
"""Angular distance between the center of the region and the given position.
Parameters
----------
position : `astropy.coordinates.SkyCoord`
Sky coordinate we want the angular distance to.
Returns
-------
sep : `~astropy.coordinates.Angle`
The on-sky separation between the given coordinate and the region center.
"""
return self.center_skydir.separation(position)
@property
def data_shape(self):
"""Shape of the Numpy data array matching this geometry."""
return self._shape[::-1]
@property
def data_shape_axes(self):
"""Shape of data of the non-spatial axes and unit spatial axes."""
return self.axes.shape[::-1] + (1, 1)
@property
def _shape(self):
"""Number of bins in each dimension.
The spatial dimension is always (1, 1), as a
`RegionGeom` is not pixelized further
"""
return tuple((1, 1) + self.axes.shape)
[docs] def get_coord(self, mode="center", frame=None, sparse=False, axis_name=None):
"""Get map coordinates from the geometry.
Parameters
----------
mode : {'center', 'edges'}
Get center or edge coordinates for the non-spatial axes.
frame : str or `~astropy.coordinates.Frame`
Coordinate frame
sparse : bool
Compute sparse coordinates
axis_name : str
If mode = "edges", the edges will be returned for this axis only.
Returns
-------
coord : `~MapCoord`
Map coordinate object.
"""
if mode == "edges" and axis_name is None:
raise ValueError("Mode 'edges' requires axis name")
coords = self.axes.get_coord(mode=mode, axis_name=axis_name)
coords["skycoord"] = self.center_skydir.reshape((1, 1))
if frame is None:
frame = self.frame
return MapCoord.create(coords, frame=self.frame).to_frame(frame)
def _pad_spatial(self, pad_width):
raise NotImplementedError("Spatial padding of `RegionGeom` not supported")
[docs] def crop(self):
raise NotImplementedError("Cropping of `RegionGeom` not supported")
[docs] def solid_angle(self):
"""Get solid angle of the region.
Returns
-------
angle : `~astropy.units.Quantity`
Solid angle of the region. In sr.
Units: ``sr``
"""
if self.region is None:
raise ValueError("Region definition required.")
# compound regions do not implement area()
# so we use the mask represenation and estimate the area
# from the pixels in the mask using oversampling
if isinstance(self.region, CompoundSkyRegion):
# oversample by a factor of ten
oversampling = 10.0
wcs = self.to_binsz_wcs(self.binsz_wcs / oversampling).wcs
pixel_region = self.region.to_pixel(wcs)
mask = pixel_region.to_mask()
area = np.count_nonzero(mask) / oversampling**2
else:
# all other types of regions should implement area
area = self.region.to_pixel(self.wcs).area
solid_angle = area * proj_plane_pixel_area(self.wcs) * u.deg**2
return solid_angle.to("sr")
[docs] def bin_volume(self):
"""If the RegionGeom has a non-spatial axis, it
returns the volume of the region. If not, it
just returns the solid angle size.
Returns
-------
volume : `~astropy.units.Quantity`
Volume of the region.
"""
bin_volume = self.solid_angle() * np.ones(self.data_shape)
for idx, ax in enumerate(self.axes):
shape = self.ndim * [1]
shape[-(idx + 3)] = -1
bin_volume = bin_volume * ax.bin_width.reshape(tuple(shape))
return bin_volume
[docs] def to_wcs_geom(self, width_min=None):
"""Get the minimal equivalent geometry
which contains the region.
Parameters
----------
width_min : `~astropy.quantity.Quantity`
Minimal width for the resulting geometry.
Can be a single number or two, for
different minimum widths in each spatial dimension.
Returns
-------
wcs_geom : `~WcsGeom`
A WCS geometry object.
"""
if width_min is not None:
width = np.max(
[self.width.to_value("deg"), _check_width(width_min)], axis=0
)
else:
width = self.width
wcs_geom_region = WcsGeom(wcs=self.wcs, npix=self.wcs.array_shape)
wcs_geom = wcs_geom_region.cutout(position=self.center_skydir, width=width)
wcs_geom = wcs_geom.to_cube(self.axes)
return wcs_geom
[docs] def to_binsz_wcs(self, binsz):
"""Change the bin size of the underlying WCS geometry.
Parameters
----------
binzs : float, string or `~astropy.quantity.Quantity`
Returns
-------
region : `~RegionGeom`
A RegionGeom with the same axes and region as the input,
but different wcs pixelization.
"""
new_geom = RegionGeom(self.region, axes=self.axes, binsz_wcs=binsz)
return new_geom
[docs] def get_wcs_coord_and_weights(self, factor=10):
"""Get the array of spatial coordinates and corresponding weights
The coordinates are the center of a pixel that intersects the region and
the weights that represent which fraction of the pixel is contained
in the region.
Parameters
----------
factor : int
Oversampling factor to compute the weights
Returns
-------
region_coord : `~MapCoord`
MapCoord object with the coordinates inside
the region.
weights : `~np.array`
Weights representing the fraction of each pixel
contained in the region.
"""
wcs_geom = self.to_wcs_geom()
weights = wcs_geom.to_image().region_weights(
regions=[self.region], oversampling_factor=factor
)
mask = weights.data > 0
weights = weights.data[mask]
# Get coordinates
coords = wcs_geom.get_coord(sparse=True).apply_mask(mask)
return coords, weights
[docs] def to_binsz(self, binsz):
"""Returns self"""
return self
[docs] def to_cube(self, axes):
"""Append non-spatial axes to create a higher-dimensional geometry.
Returns
-------
region : `~RegionGeom`
RegionGeom with the added axes.
"""
axes = copy.deepcopy(self.axes) + axes
return self._init_copy(axes=axes)
[docs] def to_image(self):
"""Remove non-spatial axes to create a 2D region.
Returns
-------
region : `~RegionGeom`
RegionGeom without any non-spatial axes.
"""
return self._init_copy(axes=None)
[docs] def upsample(self, factor, axis_name=None):
"""Upsample a non-spatial dimension of the region by a given factor.
Returns
-------
region : `~RegionGeom`
RegionGeom with the upsampled axis.
"""
axes = self.axes.upsample(factor=factor, axis_name=axis_name)
return self._init_copy(axes=axes)
[docs] def downsample(self, factor, axis_name):
"""Downsample a non-spatial dimension of the region by a given factor.
Returns
-------
region : `~RegionGeom`
RegionGeom with the downsampled axis.
"""
axes = self.axes.downsample(factor=factor, axis_name=axis_name)
return self._init_copy(axes=axes)
[docs] def pix_to_coord(self, pix):
lon = np.where(
(-0.5 < pix[0]) & (pix[0] < 0.5),
self.center_skydir.data.lon,
np.nan * u.deg,
)
lat = np.where(
(-0.5 < pix[1]) & (pix[1] < 0.5),
self.center_skydir.data.lat,
np.nan * u.deg,
)
coords = (lon, lat)
for p, ax in zip(pix[self._slice_non_spatial_axes], self.axes):
coords += (ax.pix_to_coord(p),)
return coords
[docs] def pix_to_idx(self, pix, clip=False):
idxs = list(pix_tuple_to_idx(pix))
for i, idx in enumerate(idxs[self._slice_non_spatial_axes]):
if clip:
np.clip(idx, 0, self.axes[i].nbin - 1, out=idx)
else:
np.putmask(idx, (idx < 0) | (idx >= self.axes[i].nbin), -1)
return tuple(idxs)
[docs] def coord_to_pix(self, coords):
coords = MapCoord.create(coords, frame=self.frame, axis_names=self.axes.names)
if self.region is None:
pix = (0, 0)
else:
in_region = self.region.contains(coords.skycoord, wcs=self.wcs)
x = np.zeros(coords.skycoord.shape)
x[~in_region] = np.nan
y = np.zeros(coords.skycoord.shape)
y[~in_region] = np.nan
pix = (x, y)
pix += self.axes.coord_to_pix(coords)
return pix
[docs] def get_idx(self):
idxs = [np.arange(n, dtype=float) for n in self.data_shape[::-1]]
return np.meshgrid(*idxs[::-1], indexing="ij")[::-1]
def _make_bands_cols(self):
return []
[docs] @classmethod
def create(cls, region, **kwargs):
"""Create region geometry.
The input region can be passed in the form of a ds9 string and will be parsed
internally by `~regions.Regions.parse`. See:
* https://astropy-regions.readthedocs.io/en/stable/region_io.html
* http://ds9.si.edu/doc/ref/region.html
Parameters
----------
region : str or `~regions.SkyRegion`
Region definition
**kwargs : dict
Keyword arguments passed to `RegionGeom.__init__`
Returns
-------
geom : `RegionGeom`
Region geometry
"""
return cls.from_regions(regions=region, **kwargs)
def __repr__(self):
axes = ["lon", "lat"] + [_.name for _ in self.axes]
try:
frame = self.center_skydir.frame.name
lon = self.center_skydir.data.lon.deg
lat = self.center_skydir.data.lat.deg
except AttributeError:
frame, lon, lat = "", np.nan, np.nan
return (
f"{self.__class__.__name__}\n\n"
f"\tregion : {self.region.__class__.__name__}\n"
f"\taxes : {axes}\n"
f"\tshape : {self.data_shape[::-1]}\n"
f"\tndim : {self.ndim}\n"
f"\tframe : {frame}\n"
f"\tcenter : {lon:.1f} deg, {lat:.1f} deg\n"
)
[docs] def is_allclose(self, other, rtol_axes=1e-6, atol_axes=1e-6):
"""Compare two data IRFs for equivalency
Parameters
----------
other : `RegionGeom`
Geom to compare against.
rtol_axes : float
Relative tolerance for the axes comparison.
atol_axes : float
Relative tolerance for the axes comparison.
Returns
-------
is_allclose : bool
Whether the geometry is all close.
"""
if not isinstance(other, self.__class__):
return TypeError(f"Cannot compare {type(self)} and {type(other)}")
if self.data_shape != other.data_shape:
return False
axes_eq = self.axes.is_allclose(other.axes, rtol=rtol_axes, atol=atol_axes)
# TODO: compare regions based on masks...
regions_eq = True
return axes_eq and regions_eq
def __eq__(self, other):
if not isinstance(other, self.__class__):
return False
return self.is_allclose(other=other)
def _to_region_table(self):
"""Export region to a FITS region table."""
if self.region is None:
raise ValueError("Region definition required.")
region_list = compound_region_to_regions(self.region)
pixel_region_list = []
for reg in region_list:
pixel_region_list.append(reg.to_pixel(self.wcs))
table = Regions(pixel_region_list).serialize(format="fits")
header = WcsGeom(wcs=self.wcs, npix=self.wcs.array_shape).to_header()
table.meta.update(header)
return table
[docs] def to_hdulist(self, format="ogip", hdu_bands=None, hdu_region=None):
"""Convert geom to hdulist
Parameters
----------
format : {"gadf", "ogip", "ogip-sherpa"}
HDU format
hdu : str
Name of the HDU with the map data.
Returns
-------
hdulist : `~astropy.io.fits.HDUList`
HDU list
"""
if hdu_bands is None:
hdu_bands = "HDU_BANDS"
if hdu_region is None:
hdu_region = "HDU_REGION"
if format != "gadf":
hdu_region = "REGION"
hdulist = fits.HDUList()
hdulist.append(self.axes.to_table_hdu(hdu_bands=hdu_bands, format=format))
# region HDU
if self.region:
region_table = self._to_region_table()
region_hdu = fits.BinTableHDU(region_table, name=hdu_region)
hdulist.append(region_hdu)
return hdulist
[docs] @classmethod
def from_regions(cls, regions, **kwargs):
"""Create region geom from list of regions
The regions are combined with union to a compound region.
Parameters
----------
regions : list of `~regions.SkyRegion` or str
Regions
**kwargs: dict
Keyword arguments forwarded to `RegionGeom`
Returns
-------
geom : `RegionGeom`
Region map geometry
"""
if isinstance(regions, str):
regions = Regions.parse(data=regions, format="ds9")
elif isinstance(regions, SkyRegion):
regions = [regions]
elif isinstance(regions, SkyCoord):
regions = [PointSkyRegion(center=regions)]
if regions:
regions = regions_to_compound_region(regions)
return cls(region=regions, **kwargs)
[docs] @classmethod
def from_hdulist(cls, hdulist, format="ogip", hdu=None):
"""Read region table and convert it to region list.
Parameters
----------
hdulist : `~astropy.io.fits.HDUList`
HDU list
format : {"ogip", "ogip-arf", "gadf"}
HDU format
Returns
-------
geom : `RegionGeom`
Region map geometry
"""
region_hdu = "REGION"
if format == "gadf" and hdu:
region_hdu = hdu + "_" + region_hdu
if region_hdu in hdulist:
try:
region_table = QTable.read(hdulist[region_hdu])
regions_pix = Regions.parse(data=region_table, format="fits")
except TypeError:
# TODO: this is needed to support regions=0.5
region_table = Table.read(hdulist[region_hdu])
regions_pix = Regions.parse(data=region_table, format="fits")
wcs = WcsGeom.from_header(region_table.meta).wcs
regions = []
for region_pix in regions_pix:
# TODO: remove workaround once regions issue with fits serialization is sorted out
# see https://github.com/astropy/regions/issues/400
region_pix.meta["include"] = True
regions.append(region_pix.to_sky(wcs))
region = regions_to_compound_region(regions)
else:
region, wcs = None, None
if format == "ogip":
hdu_bands = "EBOUNDS"
elif format == "ogip-arf":
hdu_bands = "SPECRESP"
elif format == "gadf":
hdu_bands = hdu + "_BANDS"
else:
raise ValueError(f"Unknown format {format}")
axes = MapAxes.from_table_hdu(hdulist[hdu_bands], format=format)
return cls(region=region, wcs=wcs, axes=axes)
[docs] def union(self, other):
"""Stack a RegionGeom by making the union"""
if not self == other:
raise ValueError("Can only make union if extra axes are equivalent.")
if other.region:
if self.region:
self._region = self.region.union(other.region)
else:
self._region = other.region
[docs] def plot_region(self, ax=None, kwargs_point=None, path_effect=None, **kwargs):
"""Plot region in the sky.
Parameters
----------
ax : `~astropy.visualization.WCSAxes`
Axes to plot on. If no axes are given,
the region is shown using the minimal
equivalent WCS geometry.
kwargs_point : dict
Keyword arguments passed to `~matplotlib.lines.Line2D` for plotting
of point sources
path_effect : `~matplotlib.patheffects.PathEffect`
Path effect applied to artists and lines.
**kwargs : dict
Keyword arguments forwarded to `~regions.PixelRegion.as_artist`
Returns
-------
ax : `~astropy.visualization.WCSAxes`
Axes to plot on.
"""
from astropy.visualization.wcsaxes import WCSAxes
kwargs_point = kwargs_point or {}
if ax is None:
ax = plt.gca()
if not isinstance(ax, WCSAxes):
ax.remove()
wcs_geom = self.to_wcs_geom()
m = Map.from_geom(wcs_geom.to_image())
ax = m.plot(add_cbar=False)
kwargs.setdefault("fc", "None")
kwargs.setdefault("ec", "tab:blue")
kwargs_point.setdefault("color", kwargs.get("edgecolor"))
kwargs_point.setdefault("marker", "*")
for region in compound_region_to_regions(self.region):
region_pix = region.to_pixel(wcs=ax.wcs)
if isinstance(region, PointSkyRegion):
artist = region_pix.as_artist(**kwargs_point)
else:
artist = region_pix.as_artist(**kwargs)
if path_effect:
artist.add_path_effect(path_effect)
ax.add_artist(artist)
return ax