def MappedBoundingBox(transforms):
'''Calculate the bounding box of the warped position for a set of transforms'''
if len(transforms) == 1:
# Copy the data instead of passing the transforms object
return nornir_imageregistration.Rectangle(transforms[0].MappedBoundingBox.ToTuple())
mbb = None
for t in transforms:
if mbb is None:
mbb = t.MappedBoundingBox.ToArray()
else:
mbb = np.vstack((mbb, t.MappedBoundingBox.ToArray()))
minX = np.min(mbb[:, 1])
minY = np.min(mbb[:, 0])
maxX = np.max(mbb[:, 3])
maxY = np.max(mbb[:, 2])
return nornir_imageregistration.Rectangle((float(minY), float(minX), float(maxY), float(maxX)))
def test_tile_overlaps_diagonal(self):
A_fixed_center = (10, 40)
B_fixed_center = (-10, -40)
A_shape = (100, 100)
B_shape = (100, 100)
tile_A = self.create_tile(A_fixed_center, A_shape)
tile_B = self.create_tile(B_fixed_center, B_shape)
overlap = nornir_imageregistration.tile_overlap.TileOverlap(
tile_A, tile_B, 1, 1)
# plot_tile_overlap(overlap)
expected_overlap_A = nornir_imageregistration.Rectangle(
(-50, -50, 30, -30))
numpy.testing.assert_array_equal(
overlap.scaled_overlapping_source_rect_A.BoundingBox,
expected_overlap_A.BoundingBox)
expected_overlap_B = nornir_imageregistration.Rectangle(
(-30, 30, 50, 50))
numpy.testing.assert_array_equal(
overlap.scaled_overlapping_source_rect_B.BoundingBox,
expected_overlap_B.BoundingBox)
return
def CropImage(imageparam, Xo, Yo, Width, Height, cval=None):
'''
Crop the image at the passed bounds and returns the cropped ndarray.
If the requested area is outside the bounds of the array then the correct region is returned
with a background color set
:param ndarray imageparam: An ndarray image to crop. A string containing a path to an image is also acceptable.e
:param int Xo: X origin for crop
:param int Yo: Y origin for crop
:param int Width: New width of image
:param int Height: New height of image
:param int cval: default value for regions outside the original image boundaries. Defaults to 0. Use 'random' to fill with random noise matching images statistical profile
:return: Cropped image
:rtype: ndarray
'''
image = ImageParamToImageArray(imageparam)
if image is None:
return None
# if not isinstance(Width, int):
# Width = int(Width)
#
# if not isinstance(Height, int):
# Height = int(Height)
assert(isinstance(Width, int))
assert(isinstance(Height, int))
if Width < 0:
raise ValueError("Negative dimensions are not allowed")
if Height < 0:
raise ValueError("Negative dimensions are not allowed")
image_rectangle = nornir_imageregistration.Rectangle([0, 0, image.shape[0], image.shape[1]])
crop_rectangle = nornir_imageregistration.Rectangle.CreateFromPointAndArea([Yo, Xo], [Height, Width])
overlap_rectangle = nornir_imageregistration.Rectangle.overlap_rect(image_rectangle, crop_rectangle)
in_startY = Yo
in_startX = Xo
in_endX = Xo + Width
in_endY = Yo + Height
out_startY = 0
out_startX = 0
out_endX = Width
out_endY = Height
if overlap_rectangle is None:
out_startY = 0
out_startX = 0
out_endX = 0
out_endY = 0
in_startY = Yo
in_startX = Xo
in_endX = Xo
in_endY = Yo
else:
(in_startY, in_startX) = overlap_rectangle.BottomLeft
(in_endY, in_endX) = overlap_rectangle.TopRight
(out_startY, out_startX) = overlap_rectangle.BottomLeft - crop_rectangle.BottomLeft
(out_endY, out_endX) = np.array([out_startY, out_startX]) + overlap_rectangle.Size
#To correct a numpy warning, convert values to int
in_startX = int(in_startX)
in_startY = int(in_startY)
in_endX = int(in_endX)
in_endY = int(in_endY)
out_startX = int(out_startX)
out_startY = int(out_startY)
out_endX = int(out_endX)
out_endY = int(out_endY)
# Create mask
rMask = None
if cval == 'random':
rMask = np.zeros((Height, Width), dtype=np.bool)
rMask[out_startY:out_endY, out_startX:out_endX] = True
# Create output image
cropped = None
if cval is None:
cropped = np.zeros((Height, Width), dtype=image.dtype)
elif cval == 'random':
cropped = np.ones((Height, Width), dtype=image.dtype)
else:
cropped = np.ones((Height, Width), dtype=image.dtype) * cval
cropped[out_startY:out_endY, out_startX:out_endX] = image[in_startY:in_endY, in_startX:in_endX]
if not rMask is None:
return RandomNoiseMask(cropped, rMask, Copy=False)
return cropped