BoundingBox¶

class regions.BoundingBox(ixmin, ixmax, iymin, iymax)[source]

Bases: object

A rectangular bounding box in integer (not float) pixel indices.

Parameters
ixmin, ixmax, iymin, iymaxint

The bounding box pixel indices. Note that the upper values (iymax and ixmax) are exclusive as for normal slices in Python. The lower values (ixmin and iymin) must not be greater than the respective upper values (ixmax and iymax).

Examples

>>> from regions import BoundingBox
>>> # constructing a BoundingBox like this is cryptic:
>>> bbox = BoundingBox(1, 10, 2, 20)
>>> # it's better to use keyword arguments for readability:
>>> bbox = BoundingBox(ixmin=1, ixmax=10, iymin=2, iymax=20)
>>> bbox  # nice repr, useful for interactive work
BoundingBox(ixmin=1, ixmax=10, iymin=2, iymax=20)
>>> # sometimes it's useful to check if two bounding boxes are the same
>>> bbox == BoundingBox(ixmin=1, ixmax=10, iymin=2, iymax=20)
True
>>> bbox == BoundingBox(ixmin=7, ixmax=10, iymin=2, iymax=20)
False
>>> # "center" and "shape" can be useful when working with numpy arrays
>>> bbox.center  # numpy order: (y, x)
(10.5, 5.0)
>>> bbox.shape  # numpy order: (y, x)
(18, 9)
>>> # "extent" is useful when plotting the BoundingBox with matplotlib
>>> bbox.extent  # matplotlib order: (x, y)
(0.5, 9.5, 1.5, 19.5)

Attributes Summary

 center The (y, x) center of the bounding box. extent The extent of the mask, defined as the (xmin, xmax, ymin, ymax) bounding box from the bottom-left corner of the lower-left pixel to the upper-right corner of the upper-right pixel. shape The (ny, nx) shape of the bounding box. slices Deprecated since version 0.5.

Methods Summary

 as_artist(**kwargs) Return a matplotlib.patches.Rectangle that represents the bounding box. from_float(xmin, xmax, ymin, ymax) Return the smallest bounding box that fully contains a given rectangle defined by float coordinate values. get_overlap_slices(shape) Get slices for the overlapping part of the bounding box and an 2D array. intersection(other) Return a BoundingBox representing the intersection of this BoundingBox with another BoundingBox. plot([origin, ax]) Plot the BoundingBox on a matplotlib Axes instance. Return a RectanglePixelRegion that represents the bounding box. union(other) Return a BoundingBox representing the union of this BoundingBox with another BoundingBox.

Attributes Documentation

center

The (y, x) center of the bounding box.

extent

The extent of the mask, defined as the (xmin, xmax, ymin, ymax) bounding box from the bottom-left corner of the lower-left pixel to the upper-right corner of the upper-right pixel.

The upper edges here are the actual pixel positions of the edges, i.e., they are not “exclusive” indices used for python indexing. This is useful for plotting the bounding box using Matplotlib.

shape

The (ny, nx) shape of the bounding box.

slices

Deprecated since version 0.5: The slices function is deprecated and may be removed in a future version. Use get_overlap_slices instead.

The bounding box as a tuple of slice objects.

The slice tuple is in numpy axis order (i.e., (y, x)) and therefore can be used to slice numpy arrays.

Methods Documentation

as_artist(**kwargs)[source]

Return a matplotlib.patches.Rectangle that represents the bounding box.

Parameters
**kwargsdict

Any keyword arguments accepted by matplotlib.patches.Patch.

Returns
resultmatplotlib.patches.Rectangle

A matplotlib rectangular patch.

Examples

import numpy as np
import matplotlib.pyplot as plt
from regions import BoundingBox
bbox = BoundingBox(2, 7, 3, 8)
fig = plt.figure()
rng = np.random.default_rng(0)
ax.imshow(rng.random((10, 10)), interpolation='nearest',
cmap='viridis')
lw=2.))
classmethod from_float(xmin, xmax, ymin, ymax)[source]

Return the smallest bounding box that fully contains a given rectangle defined by float coordinate values.

Following the pixel index convention, an integer index corresponds to the center of a pixel and the pixel edges span from (index - 0.5) to (index + 0.5). For example, the pixel edge spans of the following pixels are:

• pixel 0: from -0.5 to 0.5

• pixel 1: from 0.5 to 1.5

• pixel 2: from 1.5 to 2.5

In addition, because BoundingBox upper limits are exclusive (by definition), 1 is added to the upper pixel edges. See examples below.

Parameters
xmin, xmax, ymin, ymaxfloat

Float coordinates defining a rectangle. The lower values (xmin and ymin) must not be greater than the respective upper values (xmax and ymax).

Returns
bboxBoundingBox object

The minimal BoundingBox object fully containing the input rectangle coordinates.

Examples

>>> from regions import BoundingBox
>>> BoundingBox.from_float(xmin=1.0, xmax=10.0, ymin=2.0, ymax=20.0)
BoundingBox(ixmin=1, ixmax=11, iymin=2, iymax=21)
>>> BoundingBox.from_float(xmin=1.4, xmax=10.4, ymin=1.6, ymax=10.6)
BoundingBox(ixmin=1, ixmax=11, iymin=2, iymax=12)
get_overlap_slices(shape)[source]

Get slices for the overlapping part of the bounding box and an 2D array.

Parameters
shape2-tuple of int

The shape of the 2D array.

Returns
slices_largetuple of slices or None

A tuple of slice objects for each axis of the large array, such that large_array[slices_large] extracts the region of the large array that overlaps with the small array. None is returned if there is no overlap of the bounding box with the given image shape.

slices_smalltuple of slices or None

A tuple of slice objects for each axis of an array enclosed by the bounding box such that small_array[slices_small] extracts the region that is inside the large array. None is returned if there is no overlap of the bounding box with the given image shape.

intersection(other)[source]

Return a BoundingBox representing the intersection of this BoundingBox with another BoundingBox.

Parameters
otherBoundingBox

The BoundingBox to intersect with this one.

Returns
resultBoundingBox

A BoundingBox representing the intersection of the input BoundingBox with this one.

plot(origin=(0, 0), ax=None, **kwargs)[source]

Plot the BoundingBox on a matplotlib Axes instance.

Parameters
originarray_like, optional

The (x, y) position of the origin of the displayed image.

axmatplotlib.axes.Axes, optional

If None, then the current Axes instance is used.

**kwargsdict

Any keyword arguments accepted by matplotlib.patches.Patch.

Returns
axAxes

Axes on which the patch is added.

to_region()[source]

Return a RectanglePixelRegion that represents the bounding box.

union(other)[source]

Return a BoundingBox representing the union of this BoundingBox with another BoundingBox.

Parameters
otherBoundingBox

The BoundingBox to join with this one.

Returns
resultBoundingBox

A BoundingBox representing the union of the input BoundingBox with this one.