def cooccurrence(quantized_image, labels, scale_i=3, scale_j=0):
"""Calculates co-occurrence matrices for all the objects in the image.
Return an array P of shape (nobjects, nlevels, nlevels) such that
P[o, :, :] is the cooccurence matrix for object o.
quantized_image -- a numpy array of integer type
labels -- a numpy array of integer type
scale -- an integer
For each object O, the cooccurrence matrix is defined as follows.
Given a row number I in the matrix, let A be the set of pixels in
O with gray level I, excluding pixels in the rightmost S
columns of the image. Let B be the set of pixels in O that are S
pixels to the right of a pixel in A. Row I of the cooccurence
matrix is the gray-level histogram of the pixels in B.
"""
labels = labels.astype(int)
nlevels = quantized_image.max() + 1
nobjects = labels.max()
if scale_i < 0:
scale_i = -scale_i
scale_j = -scale_j
if scale_i == 0 and scale_j > 0:
image_a = quantized_image[:, :-scale_j]
image_b = quantized_image[:, scale_j:]
labels_ab = labels_a = labels[:, :-scale_j]
labels_b = labels[:, scale_j:]
elif scale_i > 0 and scale_j == 0:
image_a = quantized_image[:-scale_i, :]
image_b = quantized_image[scale_i:, :]
labels_ab = labels_a = labels[:-scale_i, :]
labels_b = labels[scale_i:, :]
elif scale_i > 0 and scale_j > 0:
image_a = quantized_image[:-scale_i, :-scale_j]
image_b = quantized_image[scale_i:, scale_j:]
labels_ab = labels_a = labels[:-scale_i, :-scale_j]
labels_b = labels[scale_i:, scale_j:]
else:
# scale_j should be negative
image_a = quantized_image[:-scale_i, -scale_j:]
image_b = quantized_image[scale_i:, :scale_j]
labels_ab = labels_a = labels[:-scale_i, -scale_j:]
labels_b = labels[scale_i:, :scale_j]
equilabel = (labels_a == labels_b) & (labels_a > 0)
if np.any(equilabel):
Q = nlevels * nlevels * (labels_ab[equilabel] - 1) + nlevels * image_a[equilabel] + image_b[equilabel]
R = np.bincount(Q)
if R.size != nobjects * nlevels * nlevels:
S = np.zeros(nobjects * nlevels * nlevels - R.size)
R = np.hstack((R, S))
P = R.reshape(nobjects, nlevels, nlevels)
pixel_count = fix(scind.sum(equilabel, labels_ab, np.arange(nobjects, dtype=np.int32) + 1))
pixel_count = np.tile(pixel_count[:, np.newaxis, np.newaxis], (1, nlevels, nlevels))
return (P.astype(float) / pixel_count.astype(float), nlevels)
else:
return np.zeros((nobjects, nlevels, nlevels)), nlevels
def cooccurrence(quantized_image, labels, scale_i=3, scale_j=0):
"""Calculates co-occurrence matrices for all the objects in the image.
Return an array P of shape (nobjects, nlevels, nlevels) such that
P[o, :, :] is the cooccurence matrix for object o.
quantized_image -- a numpy array of integer type
labels -- a numpy array of integer type
scale -- an integer
For each object O, the cooccurrence matrix is defined as follows.
Given a row number I in the matrix, let A be the set of pixels in
O with gray level I, excluding pixels in the rightmost S
columns of the image. Let B be the set of pixels in O that are S
pixels to the right of a pixel in A. Row I of the cooccurence
matrix is the gray-level histogram of the pixels in B.
"""
labels = labels.astype(int)
nlevels = quantized_image.max() + 1
nobjects = labels.max()
if scale_i < 0:
scale_i = -scale_i
scale_j = -scale_j
if scale_i == 0 and scale_j > 0:
image_a = quantized_image[:, :-scale_j]
image_b = quantized_image[:, scale_j:]
labels_ab = labels_a = labels[:, :-scale_j]
labels_b = labels[:, scale_j:]
elif scale_i > 0 and scale_j == 0:
image_a = quantized_image[:-scale_i, :]
image_b = quantized_image[scale_i:, :]
labels_ab = labels_a = labels[:-scale_i, :]
labels_b = labels[scale_i:, :]
elif scale_i > 0 and scale_j > 0:
image_a = quantized_image[:-scale_i, :-scale_j]
image_b = quantized_image[scale_i:, scale_j:]
labels_ab = labels_a = labels[:-scale_i, :-scale_j]
labels_b = labels[scale_i:, scale_j:]
else:
# scale_j should be negative
image_a = quantized_image[:-scale_i, -scale_j:]
image_b = quantized_image[scale_i:, :scale_j]
labels_ab = labels_a = labels[:-scale_i, -scale_j:]
labels_b = labels[scale_i:, :scale_j]
equilabel = ((labels_a == labels_b) & (labels_a > 0))
if np.any(equilabel):
Q = (nlevels * nlevels * (labels_ab[equilabel] - 1) +
nlevels * image_a[equilabel] + image_b[equilabel])
R = np.bincount(Q)
if R.size != nobjects * nlevels * nlevels:
S = np.zeros(nobjects * nlevels * nlevels - R.size)
R = np.hstack((R, S))
P = R.reshape(nobjects, nlevels, nlevels)
pixel_count = fix(
scind.sum(equilabel, labels_ab,
np.arange(nobjects, dtype=np.int32) + 1))
pixel_count = np.tile(pixel_count[:, np.newaxis, np.newaxis],
(1, nlevels, nlevels))
return (P.astype(float) / pixel_count.astype(float), nlevels)
else:
return np.zeros((nobjects, nlevels, nlevels)), nlevels
def maximum(input, labels, index):
return fix(scind.maximum(input, labels, index))
labels_b = labels[scale_i:, :]
elif scale_i > 0 and scale_j > 0:
image_a = quantized_image[:-scale_i, :-scale_j]
image_b = quantized_image[scale_i:, scale_j:]
labels_ab = labels_a = labels[:-scale_i, :-scale_j]
labels_b = labels[scale_i:, scale_j:]
else:
# scale_j should be negative
image_a = quantized_image[:-scale_i, -scale_j:]
image_b = quantized_image[scale_i:, :scale_j]
labels_ab = labels_a = labels[:-scale_i, -scale_j:]
labels_b = labels[scale_i:, :scale_j]
equilabel = ((labels_a == labels_b) & (labels_a > 0))
if np.any(equilabel):
Q = (nlevels*nlevels*(labels_ab[equilabel]-1)+
nlevels*image_a[equilabel]+image_b[equilabel])
R = np.bincount(Q)
if R.size != nobjects*nlevels*nlevels:
S = np.zeros(nobjects*nlevels*nlevels-R.size)
R = np.hstack((R, S))
P = R.reshape(nobjects, nlevels, nlevels)
pixel_count = fix(scind.sum(equilabel, labels_ab,
np.arange(nobjects, dtype=np.int32)+1))
pixel_count = np.tile(pixel_count[:,np.newaxis,np.newaxis],
(1,nlevels,nlevels))
return (P.astype(float) / pixel_count.astype(float), nlevels)
else:
return np.zeros((nobjects, nlevels, nlevels)), nlevels
labels_b = labels[scale_i:, :]
elif scale_i > 0 and scale_j > 0:
image_a = quantized_image[:-scale_i, :-scale_j]
image_b = quantized_image[scale_i:, scale_j:]
labels_ab = labels_a = labels[:-scale_i, :-scale_j]
labels_b = labels[scale_i:, scale_j:]
else:
# scale_j should be negative
image_a = quantized_image[:-scale_i, -scale_j:]
image_b = quantized_image[scale_i:, :scale_j]
labels_ab = labels_a = labels[:-scale_i, -scale_j:]
labels_b = labels[scale_i:, :scale_j]
equilabel = ((labels_a == labels_b) & (labels_a > 0))
if np.any(equilabel):
Q = (nlevels * nlevels * (labels_ab[equilabel] - 1) +
nlevels * image_a[equilabel] + image_b[equilabel])
R = np.bincount(Q)
if R.size != nobjects * nlevels * nlevels:
S = np.zeros(nobjects * nlevels * nlevels - R.size)
R = np.hstack((R, S))
P = R.reshape(nobjects, nlevels, nlevels)
pixel_count = fix(
scind.sum(equilabel, labels_ab,
np.arange(nobjects, dtype=np.int32) + 1))
pixel_count = np.tile(pixel_count[:, np.newaxis, np.newaxis],
(1, nlevels, nlevels))
return (P.astype(float) / pixel_count.astype(float), nlevels)
else:
return np.zeros((nobjects, nlevels, nlevels)), nlevels
def maximum(input, labels, index):
return fix(scind.maximum(input, labels, index))
