def mainProcess(image_original, image_transformed):
global pathIn
global keypoints_original_blob
global keypoints_transformed_blob
image_gray = rgb2gray(image_transformed)
img_path_original = skimage.io.imread(pathIn)
img_original = rgb2gray(image_original)
blobs_doh = blob_doh(img_original,
min_sigma=1,
max_sigma=30,
num_sigma=10,
threshold=0.01,
overlap=0.5,
log_scale=False)
keypoints_original_blob = blobs_doh
blobs_doh2 = blob_doh(image_gray,
min_sigma=1,
max_sigma=30,
num_sigma=10,
threshold=0.01,
overlap=0.5,
log_scale=False)
keypoints_transformed_blob = blobs_doh2
return blobs_doh, blobs_doh2
def _detect_specimens(self,
reconstructed_wave,
propagation_distance,
margin=100,
kernel_radius=4.0,
save_png_to_disk=None):
cropped_img = reconstructed_wave.phase[margin:-margin, margin:-margin]
best_convolved_phase = convolve_fft(cropped_img,
MexicanHat2DKernel(kernel_radius))
best_convolved_phase_copy = best_convolved_phase.copy(order='C')
# Find positive peaks
blob_doh_kwargs = dict(threshold=0.00007, min_sigma=2, max_sigma=10)
blobs = blob_doh(best_convolved_phase_copy, **blob_doh_kwargs)
# Find negative peaks
negative_phase = -best_convolved_phase_copy
negative_phase += (np.median(best_convolved_phase_copy) -
np.median(negative_phase))
negative_blobs = blob_doh(negative_phase, **blob_doh_kwargs)
all_blobs = []
for blob in blobs:
if blob.size > 0:
all_blobs.append(blob)
for neg_blob in negative_blobs:
if neg_blob.size > 0:
all_blobs.append(neg_blob)
if len(all_blobs) > 0:
all_blobs = np.vstack(all_blobs)
# If save pngs:
if save_png_to_disk is not None:
path = "{0}/{1:.4f}.png".format(save_png_to_disk,
propagation_distance)
save_scaled_image(reconstructed_wave.phase, path, margin,
all_blobs)
# Blobs get returned in rows with [x, y, radius], so save each
# set of blobs with the propagation distance to record z
# correct blob positions for margin:
all_blobs = np.float64(all_blobs)
if len(all_blobs) > 0:
all_blobs[:, 0] += margin
all_blobs[:, 1] += margin
all_blobs[:, 2] = propagation_distance
return all_blobs
else:
return None
def _detect_specimens(self, reconstructed_wave, propagation_distance,
margin=100, kernel_radius=4.0, save_png_to_disk=None):
cropped_img = reconstructed_wave.phase[margin:-margin, margin:-margin]
best_convolved_phase = convolve_fft(cropped_img,
MexicanHat2DKernel(kernel_radius))
best_convolved_phase_copy = best_convolved_phase.copy(order='C')
# Find positive peaks
blob_doh_kwargs = dict(threshold=0.00007,
min_sigma=2,
max_sigma=10)
blobs = blob_doh(best_convolved_phase_copy, **blob_doh_kwargs)
# Find negative peaks
negative_phase = -best_convolved_phase_copy
negative_phase += (np.median(best_convolved_phase_copy) -
np.median(negative_phase))
negative_blobs = blob_doh(negative_phase, **blob_doh_kwargs)
all_blobs = []
for blob in blobs:
if blob.size > 0:
all_blobs.append(blob)
for neg_blob in negative_blobs:
if neg_blob.size > 0:
all_blobs.append(neg_blob)
if len(all_blobs) > 0:
all_blobs = np.vstack(all_blobs)
# If save pngs:
if save_png_to_disk is not None:
path = "{0}/{1:.4f}.png".format(save_png_to_disk,
propagation_distance)
save_scaled_image(reconstructed_wave.phase, path, margin, all_blobs)
# Blobs get returned in rows with [x, y, radius], so save each
# set of blobs with the propagation distance to record z
# correct blob positions for margin:
all_blobs = np.float64(all_blobs)
if len(all_blobs) > 0:
all_blobs[:, 0] += margin
all_blobs[:, 1] += margin
all_blobs[:, 2] = propagation_distance
return all_blobs
else:
return None
def blob_image_multiscale2(image, type=0, scale=2):
# function that return a list of blob_coordinates, 0 = dog, 1 = doh, 2 = log
list = []
image = norm.normalize(image)
for z, slice in tqdm(enumerate(image)):
# init list of different sigma/zoom blobs
featureblobs = []
# x = 0,1,2,3,4
if scale == 2:
# for x in xrange(0,6):
# if type == 0:
# featureblobs.append(feature.blob_dog(slice, 2**x, 2**x))
# if type == 1:
# featureblobs.append(feature.blob_doh(slice, 2**x, 2**x))
# if type == 2:
# featureblobs.append(feature.blob_log(slice, 2**x, 2**x))
for x in xrange(0, 5):
if type == 0:
featureblobs.append(
feature.blob_dog(slice, 2**x, 2**(x + 1)))
if type == 1:
featureblobs.append(
feature.blob_doh(slice, 2**x, 2**(x + 1)))
if type == 2:
featureblobs.append(
feature.blob_log(slice, 2**x, 2**(x + 1), 16, .1))
else:
for x in xrange(0, 4):
if type == 0:
featureblobs.append(feature.blob_dog(slice, 3**x, 3**x))
if type == 1:
featureblobs.append(feature.blob_doh(slice, 3**x, 3**x))
if type == 2:
featureblobs.append(feature.blob_log(slice, 3**x, 3**x))
# init list of blob coords
blob_coords = []
#print featureblobs
# start at biggest blob size
for featureblob in reversed(featureblobs):
# for every blob found of a blobsize
for blob in enumerate(featureblob):
# if that blob is not within range of another blob, add it
blob = blob[1]
if not within_range(blob, blob_coords):
blob_coords.append([z, blob[0], blob[1], blob[2]])
list.append(blob_coords)
return list
def create_blob_sequence(image):
"""
Apply multiple blob detection algorithms to compare them
:param image: image to be analysed
:return: squenece containing the blobs, colors and titles
"""
blobs_log = blob_log(image,
min_sigma=15,
max_sigma=40,
num_sigma=10,
overlap=1)
# Compute radii in the 3rd column.
blobs_log[:, 2] = blobs_log[:, 2] * sqrt(2)
blobs_dog = blob_dog(image, min_sigma=5, max_sigma=40, overlap=1)
blobs_dog[:, 2] = blobs_dog[:, 2] * sqrt(2)
blobs_doh = blob_doh(image, min_sigma=5, max_sigma=40, overlap=1)
blobs_list = [blobs_log, blobs_dog, blobs_doh]
colors = ["yellow", "green", "black", "blue"]
titles = [
"Laplacian of Gaussian",
"Difference of Gaussian",
"Determinant of Hessian",
]
sequence = zip(blobs_list, colors, titles)
return sequence
def scikit_blob(frame):
h, w = frame.shape[:2]
gray = cv2.cvtColor(frame, cv2.COLOR_BGR2GRAY)
rois = []
if config.L_GAUSSIAN_DETECTION:
blobs_log = blob_log(gray,
max_sigma=config.L_GAUSSIAN_MAX_SIGMA,
num_sigma=config.L_GAUSSIAN_NUM_SIGMA,
threshold=config.L_GAUSSIAN_THRESHOLD)
blobs_log[:, 2] = blobs_log[:, 2] * sqrt(2)
rois.extend(blobs_log)
if config.D_GAUSSIAN_DETECTION:
blobs_dog = blob_dog(gray,
max_sigma=config.D_GAUSSIAN_MAX_SIGMA,
threshold=config.D_GAUSSIAN_THRESHOLD)
blobs_dog[:, 2] = blobs_dog[:, 2] * sqrt(2)
rois.extend(blobs_dog)
if config.HESSIAN_DETECTION:
blobs_doh = blob_doh(gray,
max_sigma=config.HESSIAN_MAX_SIGMA,
min_sigma=config.HESSIAN_MIN_SIGMA,
threshold=config.HESSIAN_THRESHOLD)
rois.extend(blobs_doh)
candidates = []
for roi in rois:
(y, x, r) = roi
r = 2.5 * r
candidates.append([x, y, 2 * r, 2 * r])
return candidates
def find_blobs(filename):
feature = ""
raw_image = io.imread(filename)
for channel in range(0, 4):
if channel < 3:
image = raw_image[:,:,channel]
image_gray = rgb2gray(image)
# Smoothing
image_gray = img_as_ubyte(image_gray)
image_gray = mean_bilateral(image_gray.astype(numpy.uint16), disk(20), s0=10, s1=10)
# Increase contrast
image_gray = exposure.equalize_adapthist(image_gray, clip_limit=0.03)
# Find blobs
blobs_doh = blob_doh(image_gray, min_sigma=1, max_sigma=20, threshold=.005)
count = 0
for blob in blobs_doh:
y, x, r = blob
if (x-400)**2 + (y-400)**2 > distance:
continue
count = count + 1
feature = feature + " " + str(channel + 1) + ":" + str(count)
return feature
def main():
numberOfImages = 11
# TODO: AUTOMATICALLY GET NUMBER OF IMAGES
# Get number of images. Remeber to divide by 2 as for every relevant image,
# theres also the comparison image.
# if ".DS_Store" in os.listdir("Wheat_ROIs"):
# numberOfImages = (len(os.listdir("Wheat_ROIs")) - 1)/2;
# else:
# numberOfImages = len(os.listdir("Wheat_ROIs"))/2;
# For each ROI image in folder
for i in tqdm.tqdm(range(1, numberOfImages + 1)):
# Load image
filename = "../Wheat_ROIs/{:03d}_ROI.png".format(i)
img = misc.imread(filename)
img_gray = rgb2gray(img)
# Detect blobs. See http://scikit-image.org/docs/dev/api/skimage.feature.html#skimage.feature.blob_doh
# for function documentation
blobs = blob_doh(img_gray, min_sigma=1, max_sigma=100, threshold=.01)
# Display blobs on image and save image
fig, ax = plt.subplots()
plt.title("Number of Blobs Detected: {}".format(blobs.shape[0]))
plt.grid(False)
ax.imshow(img, interpolation='nearest')
for blob in blobs:
y, x, r = blob
c = plt.Circle((x, y), r, color='red', linewidth=2, fill=False)
ax.add_patch(c)
fig.savefig("../Wheat_ROIs/{:03d}_Blob.png".format(i))
def blob_detections(image):
image_gray = rgb2gray(image).astype('float64')/255
blobs_log = blob_log(1-image_gray, max_sigma=20, num_sigma=10, threshold=0.15)
# Compute radii in the 3rd column.
blobs_log[:, 2] = blobs_log[:, 2] * sqrt(2)
blobs_dog = blob_dog(1-image_gray, max_sigma=20, threshold=0.5)
blobs_dog[:, 2] = blobs_dog[:, 2] * sqrt(2)
blobs_doh = blob_doh(1-image_gray, max_sigma=20, threshold=.002)
blobs_list = [blobs_log, blobs_dog, blobs_doh]
colors = ['yellow', 'lime', 'red']
titles = ['Laplacian of Gaussian', 'Difference of Gaussian',
'Determinant of Hessian']
sequence = zip(blobs_list, colors, titles)
fig, axes = plt.subplots(1, 3, figsize=(9, 3), sharex=True, sharey=True,
subplot_kw={'adjustable': 'box-forced'})
ax = axes.ravel()
for idx, (blobs, color, title) in enumerate(sequence):
ax[idx].set_title(title)
ax[idx].imshow(image, interpolation='nearest', cmap='gray')
for blob in blobs:
y, x, r = blob
c = plt.Circle((x, y), r, color=color, linewidth=2, fill=False)
ax[idx].add_patch(c)
ax[idx].set_axis_off()
plt.tight_layout()
plt.show()
def hlpr(image):
image = data.hubble_deep_field()[0:500, 0:500]
image_gray = rgb2gray(image)
blobs_log = blob_log(image_gray, max_sigma=30, num_sigma=10, threshold=.1)
# Compute radii in the 3rd column.
blobs_log[:, 2] = blobs_log[:, 2] * sqrt(2)
blobs_dog = blob_dog(image_gray, max_sigma=30, threshold=.1)
blobs_dog[:, 2] = blobs_dog[:, 2] * sqrt(2)
blobs_doh = blob_doh(image_gray, max_sigma=30, threshold=.01)
blobs_list = [blobs_log, blobs_dog, blobs_doh]
colors = ['yellow', 'lime', 'red']
titles = [
'Laplacian of Gaussian', 'Difference of Gaussian',
'Determinant of Hessian'
]
sequence = zip(blobs_list, colors, titles)
for blobs, color, title in sequence:
fig, ax = plt.subplots(1, 1)
ax.set_title(title)
ax.imshow(image, interpolation='nearest')
for blob in blobs:
y, x, r = blob
c = plt.Circle((x, y), r, color=color, linewidth=2, fill=False)
ax.add_patch(c)
plt.show()
def circles(self, filename):
image = cv2.imread(filename, 0)
image_gray = rgb2gray(image)
blobs_log = blob_log(image_gray, max_sigma=30, num_sigma=10, threshold=.1)
# Compute radii in the 3rd column.
blobs_log[:, 2] = blobs_log[:, 2] * sqrt(2)
blobs_dog = blob_dog(image_gray, max_sigma=30, threshold=.1)
blobs_dog[:, 2] = blobs_dog[:, 2] * sqrt(2)
blobs_doh = blob_doh(image_gray, max_sigma=30, threshold=.01)
blobs_list = [blobs_log, blobs_doh]
colors = ['yellow', 'red']
titles = ['Laplacian of Gaussian', 'Determinant of Hessian']
sequence = zip(blobs_list, colors, titles)
fig, axes = plt.subplots(1, 2, sharex=True, sharey=True, subplot_kw={'adjustable': 'box-forced'})
axes = axes.ravel()
for blobs, color, title in sequence:
ax = axes[0]
axes = axes[1:]
ax.set_title(title)
ax.imshow(image, interpolation='nearest')
for blob in blobs:
y, x, r = blob
c = plt.Circle((x, y), r, color=color, linewidth=2, fill=False)
ax.add_patch(c)
plt.savefig('output.png')
plt.show()
def flowerCount(image_tuple):
(key, value) = image_tuple
global flower_area_mask
# Get flower mask for this image
ImageFile.LOAD_TRUNCATED_IMAGES = True
image_value = np.array(value)
img = image_value[:, :, ::-1]
# Highlight flowers
# 77 is the begining of the image name after removing file path
img_flowers = getFlowerHighlight(img, key[77:], segm_dist_from_zerocross=8)
try:
# Get number of flowers using blob counter on the B channel
blobs = blob_doh(img_flowers, max_sigma=5, min_sigma=3)
return key[77:], len(blobs), blobs
except(IndexError) as e:
pass
def test_blob_doh_2d(shape, blobs, min_sigma, max_sigma, num_sigma, overlap,
threshold, log_scale):
a = generate_blobimage(shape, blobs)
chunks = [e // 2 for e in shape]
d = da.from_array(a, chunks=chunks)
ski_r = ski_feat.blob_doh(
a,
min_sigma=min_sigma,
max_sigma=max_sigma,
num_sigma=num_sigma,
threshold=threshold,
overlap=overlap,
log_scale=log_scale,
)
da_r = da_feat.blob_doh(
d,
min_sigma=min_sigma,
max_sigma=max_sigma,
num_sigma=num_sigma,
threshold=threshold,
overlap=overlap,
log_scale=log_scale,
)
ski_r = sort_array(ski_r)
da_r = sort_array(da_r)
print(ski_r)
print(da_r)
# coordinates are sometimes off by 1 when using the determinant
coord_diff = (ski_r - da_r)[:, :a.ndim]
sigma_diff = (ski_r - da_r)[:, a.ndim:]
assert np.all(coord_diff <= 1)
assert np.all(np.abs(sigma_diff) <= 0.01)
def main():
numberOfImages = 11;
# TODO: AUTOMATICALLY GET NUMBER OF IMAGES
# Get number of images. Remeber to divide by 2 as for every relevant image,
# theres also the comparison image.
# if ".DS_Store" in os.listdir("Wheat_ROIs"):
# numberOfImages = (len(os.listdir("Wheat_ROIs")) - 1)/2;
# else:
# numberOfImages = len(os.listdir("Wheat_ROIs"))/2;
# For each ROI image in folder
for i in tqdm.tqdm(range(1, numberOfImages+1)):
# Load image
filename = "../Wheat_ROIs/{:03d}_ROI.png".format(i);
img = misc.imread(filename);
img_gray = rgb2gray(img);
# Detect blobs. See http://scikit-image.org/docs/dev/api/skimage.feature.html#skimage.feature.blob_doh
# for function documentation
blobs = blob_doh(img_gray, min_sigma=1, max_sigma=100, threshold=.01)
# Display blobs on image and save image
fig, ax = plt.subplots()
plt.title("Number of Blobs Detected: {}".format(blobs.shape[0]))
plt.grid(False)
ax.imshow(img, interpolation='nearest')
for blob in blobs:
y, x, r = blob
c = plt.Circle((x, y), r, color='red', linewidth=2, fill=False)
ax.add_patch(c)
fig.savefig("../Wheat_ROIs/{:03d}_Blob.png".format(i))
def computeFlowerCount(self, images):
print("Computing flower count")
for image in tqdm(images, file=sys.stdout):
try:
# Initialize dictionary to store number of flowers per image
n_flowers = {}
# Get flower mask for this image
PIL.ImageFile.LOAD_TRUNCATED_IMAGES = True
pil_image = PIL.Image.open(image).convert('RGB')
open_cv_image = np.array(pil_image)
img = open_cv_image[:, :, ::-1].copy()
# Highlight flowers
img_flowers = self.getFlowerHighlight(
img, image, segm_dist_from_zerocross=6)
# Apply flower area mask
img_flowers[self.flower_area_mask == 0] = 0
# Get number of flowers using blob counter on the B channel
blobs = blob_doh(img_flowers, max_sigma=5, min_sigma=3)
# Append result
n_flowers[image] = len(blobs)
except IndexError:
n_flowers[image] = []
# Save to object
self.flower_count = n_flowers
self.flower_count_estimate = n_flowers
def doh(self):
image = np.asfarray(self.image, dtype=np.double)
# skimage.feature.blob_doh(image, min_sigma=1, max_sigma=30,
# num_sigma=10, threshold=0.01, overlap=0.5, log_scale=False)
blobs = feature.blob_doh(image, **self.blob_ka, **self.doh_ka)
# Here, radius is approx. equal to sigma.
return blobs
def detect_cells(image):
image_gray = rgb2gray(image)
blobs_log = blob_log(image_gray, max_sigma=30, num_sigma=10, threshold=.1)
# Compute radii in the 3rd column.
blobs_log[:, 2] = blobs_log[:, 2] * sqrt(2)
blobs_dog = blob_dog(image_gray, max_sigma=30, threshold=.1)
blobs_dog[:, 2] = blobs_dog[:, 2] * sqrt(2)
blobs_doh = blob_doh(image_gray, max_sigma=30, threshold=.01)
blobs_list = [blobs_log, blobs_dog, blobs_doh]
colors = ['yellow', 'lime', 'red']
titles = ['Laplacian of Gaussian', 'Difference of Gaussian',
'Determinant of Hessian']
sequence = zip(blobs_list, colors, titles)
for blobs, color, title in sequence:
fig, ax = plt.subplots(1, 1)
ax.set_title(title)
ax.imshow(image, interpolation='nearest')
for blob in blobs:
y, x, r = blob
c = plt.Circle((x, y), r, color=color, linewidth=2, fill=False)
ax.add_patch(c)
plt.show()
def blob_detector(imIn, overlap, threshold, min_sigma, max_sigma, log_scale):
"""
:param imIn: Inverted Image grayscale patch
:param: Hyper-Parameters for blob detector
:overlap
:threshold
:min_sigma
:max_sigma
log_scale
:return:
"""
# Initialize and Invoke DoH blob detector
blobs_doh = blob_doh(imIn, overlap=overlap, threshold=threshold, min_sigma=min_sigma,
max_sigma=max_sigma, log_scale=log_scale)
#blobs_y, blobs_x, blobs_r = [],[],[]
blobs_y = blobs_doh[:,0]
blobs_x = blobs_doh[:,1]
blobs_r = blobs_doh[:,2]
"""for blob in blobs_doh:
y, x, r = blob # This code can be made better
blobs_y.append(y)
blobs_x.append(x)
blobs_r.append(r)"""
return blobs_y, blobs_x, blobs_r
def test_blob_overlap():
img = np.ones((256, 256), dtype=np.uint8)
xs, ys = circle(100, 100, 20)
img[xs, ys] = 255
xs, ys = circle(120, 100, 30)
img[xs, ys] = 255
blobs = blob_doh(img,
min_sigma=1,
max_sigma=60,
num_sigma=10,
threshold=.05)
assert len(blobs) == 1
r1, r2 = 7, 6
pad1, pad2 = 11, 12
blob1 = ellipsoid(r1, r1, r1)
blob1 = util.pad(blob1, pad1, mode='constant')
blob2 = ellipsoid(r2, r2, r2)
blob2 = util.pad(blob2, [(pad2, pad2), (pad2 - 9, pad2 + 9), (pad2, pad2)],
mode='constant')
im3 = np.logical_or(blob1, blob2)
blobs = blob_log(im3, min_sigma=2, max_sigma=10, overlap=0.1)
assert len(blobs) == 1
# Two circles with distance between centers equal to radius
overlap = _blob_overlap(np.array([0, 0, 10 / math.sqrt(2)]),
np.array([0, 10, 10 / math.sqrt(2)]))
assert_almost_equal(
overlap, 1. / math.pi * (2 * math.acos(1. / 2) - math.sqrt(3) / 2.))
def flowerCount(self, imageRDD):
(key, value) = imageRDD
# Trace
print "Computing flower count"
# Get flower mask for this image
ImageFile.LOAD_TRUNCATED_IMAGES = True
#pil_image = PIL.Image.open(value).convert('RGB')
open_cv_image = np.array(value)
img = open_cv_image[:, :, ::-1]
# Highlight flowers
img_flowers = self.getFlowerHighlight(img,
key,
segm_dist_from_zerocross=6)
# Apply flower area mask
img_flowers[self.flower_area_mask == 0] = 0
try:
# Get number of flowers using blob counter on the B channel
blobs = blob_doh(img_flowers, max_sigma=5, min_sigma=3)
n_blobs = len(blobs)
return key, n_blobs
except (IndexError) as e:
pass
def flowerCount(input):
(key, value) = input
#global flower_area_mask_accum
global flower_area_mask
# Get flower mask for this image
ImageFile.LOAD_TRUNCATED_IMAGES = True
open_cv_image = np.array(value)
img = open_cv_image[:, :, ::-1]
# Highlight flowers
img_flowers = getFlowerHighlight(img, key, segm_dist_from_zerocross=8)
# Apply flower area mask
img_flowers[flower_area_mask == 0] = 0
try:
# Get number of flowers using blob counter on the B channel
blobs = blob_doh(img_flowers, max_sigma=5, min_sigma=1)
n_blobs = len(blobs)
return key, n_blobs
except (IndexError) as e:
pass
def srf_detector(image):
if_detected = False
mask = create_bg1_mask(image)
image_cropped_1, bbox = crop_to_mask(image, mask)
image_cropped = rgb2gray(image_cropped_1)
image_cropped = denoise_bilateral(image_cropped,
sigma_spatial=2,
multichannel=False)
image_cropped = equalize_adapthist(image_cropped)
x = numpy.arange(image_cropped.shape[1])
first_notmask_x = [first_notmask(i, mask, bbox, image_cropped) for i in x]
plotted = ransac(x, first_notmask_x, image_cropped)
if plotted:
result = 'No SRF detected'
fig, ax = plt.subplots(1, 1, figsize=(9, 3), sharex=True, sharey=True)
ax.set_title('Cropped Image')
ax.imshow(image_cropped, cmap=plt.cm.gray, interpolation='nearest')
if plotted == False:
#only if no line can be fitted will the function try to find blobs
x_tot, y_tot = image_cropped.shape[0], image_cropped.shape[1]
blobs_doh = blob_doh(image_cropped, max_sigma=50, threshold=.0045)
plt.rcParams['image.cmap'] = 'gray'
result = 'No SRF detected'
fig, ax = plt.subplots(1, 1, figsize=(9, 3), sharex=True, sharey=True)
for blob in blobs_doh:
ax.set_title('Detected SRF in red')
ax.imshow(image_cropped, cmap=plt.cm.gray, interpolation='nearest')
y, x, r = blob
c = plt.Circle((x, y), r, color='blue', linewidth=2, fill=False)
ax.add_patch(c)
p = True
f_x = first_notmask(int(x), mask, bbox, image_cropped)
if (r > 1) and (r < 20) and (x > 30) and (x < y_tot - 30) and (
y > 50) and (y < x_tot - 50) and is_dark(
image_cropped, x,
y) and (y > f_x + (1 / 4) *
(image_cropped.shape[0] - x)):
for i in range(-int(r) - 3, int(r) + 1 + 3):
for j in range(-int(r) - 3, int(r) + 1 + 3):
if (mask[int(y) + bbox[1] + i,
int(x) + bbox[0] + j] == 0):
p = False
if p:
c = plt.Circle((x, y),
r,
color='red',
linewidth=2,
fill=False)
ax.add_patch(c)
result = 'SRF detected'
if_detected = True
print(result)
plt.show()
return if_detected
def detect_blobs(preds):
blobs = blob_doh(np.nan_to_num(preds), max_sigma=30, threshold=.05)
if blobs.shape[0] > 0:
#detected = pd.DataFrame(blobs_doh[blobs_doh[:,2] > 0], columns = ['x','y','r'])
detected = pd.DataFrame(blobs, columns=['x', 'y', 'r'])
else:
detected = pd.DataFrame(columns=['x', 'y', 'r'])
return detected
def detect_cell_boundary(img, sel_cmap):
"""Detect cell boundary and overlay the results on images, using blob
detection algorithms: Laplacian of Gaussian, Difference of Gaussian,
and Determinant of Hessian
Args:
img: input image
Returns:
blobs_list: list of detected blobs using different methods, following
this order: log, dog, and doh. Each item is a sublist of blobs,
where each sub-list item is x, y, and radius of a blob
"""
# dummy = np.copy(img)
dummy = util.invert(img)
# Laplacian of Gaussian
blobs_log = feature.blob_log(dummy, max_sigma=30, num_sigma=10,
threshold=0.01)
# Compute radii in the 3rd column.
blobs_log[:, 2] = blobs_log[:, 2] * np.sqrt(2)
# Difference of Gaussian
blobs_dog = feature.blob_dog(dummy, max_sigma=30, threshold=0.01)
blobs_dog[:, 2] = blobs_dog[:, 2] * np.sqrt(2)
# Determinant of Hessian
blobs_doh = feature.blob_doh(dummy, max_sigma=30, threshold=0.00065)
# concatenate all results
blobs_list = [blobs_log, blobs_dog, blobs_doh]
colors = ['red', 'green', 'blue']
titles = ['Laplacian of Gaussian', 'Difference of Gaussian',
'Determinant of Hessian']
# plot results
fig, axes = plt.subplots(1, 3, figsize=(12, 4), sharex=True, sharey=True)
ax = axes.ravel()
for i in range(3):
ax[i].set_title(titles[i])
ax[i].imshow(img, sel_cmap)
for blob in blobs_list[i]:
y, x, r = blob
c = plt.Circle((x, y), r, color=colors[i], linewidth=2, fill=False)
ax[i].add_patch(c)
ax[i].set_axis_off()
plt.suptitle('Detect cell boundary using blob detection')
# compute cell size
print('Average cell size (in pixel^2):')
for i in range(3):
cell_size = 0.0
for blob in blobs_list[i]:
cell_size += np.pi * blob[2]**2
if cell_size != 0.0:
cell_size /= len(blobs_list[i])
print(' - {}: {:.3f}'.format(titles[i], cell_size))
pass
def test_blob_doh(dtype, threshold_type):
img = np.ones((512, 512), dtype=dtype)
xs, ys = disk((400, 130), 20)
img[xs, ys] = 255
xs, ys = disk((460, 50), 30)
img[xs, ys] = 255
xs, ys = disk((100, 300), 40)
img[xs, ys] = 255
xs, ys = disk((200, 350), 50)
img[xs, ys] = 255
if threshold_type == 'absolute':
# Note: have to either scale up threshold or rescale the image to the
# range [0, 1] internally.
threshold = 0.05
if img.dtype.kind == 'f':
# account for lack of internal scaling to [0, 1] by img_as_float
ptp = img.ptp()
threshold *= ptp**2
threshold_rel = None
elif threshold_type == 'relative':
threshold = None
threshold_rel = 0.5
blobs = blob_doh(img,
min_sigma=1,
max_sigma=60,
num_sigma=10,
threshold=threshold,
threshold_rel=threshold_rel)
radius = lambda x: x[2]
s = sorted(blobs, key=radius)
thresh = 4
b = s[0]
assert abs(b[0] - 400) <= thresh
assert abs(b[1] - 130) <= thresh
assert abs(radius(b) - 20) <= thresh
b = s[1]
assert abs(b[0] - 460) <= thresh
assert abs(b[1] - 50) <= thresh
assert abs(radius(b) - 30) <= thresh
b = s[2]
assert abs(b[0] - 100) <= thresh
assert abs(b[1] - 300) <= thresh
assert abs(radius(b) - 40) <= thresh
b = s[3]
assert abs(b[0] - 200) <= thresh
assert abs(b[1] - 350) <= thresh
assert abs(radius(b) - 50) <= thresh
def find2Dcontours(threeDdensity, axis=0, n_cut=100):
'''
blobs are assumed to be light on dark background
each blob found, the method returns its coordinates and the standard deviation of the Gaussian kernel that detected the blob.
Parameters
----------
threeDdensity: 3D cube
density cube
axis: int
along which axis to flatten in order for skimage to find 2D contours
n_cut: float or int
lowest level from which to start looking for contours
N_cell_min: int
not implemented at the moment
overlap: float
not implemented; between 0 and 1; if the area of two blobs overlaps by a fraction > this, smaller blob is eliminated.
# min_sigma = 1 # keep it low for smaller blobs
# max_sigma = # keep it high to detect large blobs
# num_sigma =
'''
axis = int(axis)
from skimage.feature import blob_doh
from skimage.color import rgb2gray
c_min = n_cut
# flatten image along some axis
flattened_image = threeDdensity.sum(axis=axis)
flattened_image = rgb2gray(flattened_image)
# plot all contours found
fig, ax = plt.subplots()
ax.imshow(np.log10(flattened_image),
interpolation='nearest',
cmap=plt.cm.gray)
coord_sigma = blob_doh(flattened_image, threshold=n_cut)
for blob in coord_sigma:
y, x, r = blob
c = plt.Circle((x, y), r, color='r', linewidth=2, fill=False)
ax.add_patch(c)
plt.show(block=False)
# blob_log(density, min_sigma=min_sigma, max_sigma=max_sigma, )
return coord_sigma, flattened_image
def precisionRecall(self):
test_images = []
test_blobs = []
X = []
Y_ground = []
#Get blobs
for i in range(500,505):
file = self.croppedImages[i]
iter_image = rgb2gray(cv2.imread(file))
iter_blobs = blob_dog(iter_image, min_sigma=1, max_sigma=25,
sigma_ratio=1.6, threshold=.25, overlap=0.5)
test_images.append(iter_image)
test_blobs.append(iter_blobs)
print("Get blobs: " + str(i) + "/1000")
#Get X values for testing
for i in range(0, len(test_images)):
RadPic = self.RadPIC(10, test_blobs[i], rgb2gray(test_images[i]))
inputHOG = self.HOG(10, test_blobs[i], rgb2gray(test_images[i]))
X.extend(self.makeX(RadPic, inputHOG))
print("X: " + str(i) + "/1000")
#Get Y ground truth values
for i in range(500, 505):
j = i
tempStr = self.labels[j]
while tempStr.find("frame" + str(i)) == -1:
j = j + 1
tempStr = self.labels[j]
label_image = cv2.imread((glob.glob(tempStr))[0])
cropped_image = cv2.imread((glob.glob("cropped_images/frame" + str(i) + ".jpg"))[0])
cropped_image = rgb2gray(cropped_image)
blobs = blob_doh(cropped_image, min_sigma=1, max_sigma=25, num_sigma=15, threshold=.001)
for blob in blobs:
y, x, r = blob
if label_image[y,x][0] == 255 & label_image[y,x][1] == 255 & label_image[y,x][2] == 255:
Y_ground.append(1)
else:
Y_ground.append(0)
#print("Ground Truth: " + str(i) + "/1000")
#Open classifier
with open('croppedImagesV2.pkl', 'rb') as f:
forest = pickle.load(f)
#Precidicions
Y_classifier = forest.predict(X)
#Precision-Recall
precision, recall, thresholds = precision_recall_curve(Y_ground, Y_classifier)
plt.plot(recall, precision)
plt.ylabel('Precision')
plt.xlabel('Recall')
plt.show()
def blob_image_multiscale2(image, type=0,scale=2):
# function that return a list of blob_coordinates, 0 = dog, 1 = doh, 2 = log
list = []
image = norm.normalize(image)
for z, slice in tqdm(enumerate(image)):
# init list of different sigma/zoom blobs
featureblobs = []
# x = 0,1,2,3,4
if scale == 2:
# for x in xrange(0,6):
# if type == 0:
# featureblobs.append(feature.blob_dog(slice, 2**x, 2**x))
# if type == 1:
# featureblobs.append(feature.blob_doh(slice, 2**x, 2**x))
# if type == 2:
# featureblobs.append(feature.blob_log(slice, 2**x, 2**x))
for x in xrange(0,5):
if type == 0:
featureblobs.append(feature.blob_dog(slice, 2**x, 2**(x+1)))
if type == 1:
featureblobs.append(feature.blob_doh(slice, 2**x, 2**(x+1)))
if type == 2:
featureblobs.append(feature.blob_log(slice, 2**x, 2**(x+1),16,.1))
else:
for x in xrange(0,4):
if type == 0:
featureblobs.append(feature.blob_dog(slice, 3**x, 3**x))
if type == 1:
featureblobs.append(feature.blob_doh(slice, 3**x, 3**x))
if type == 2:
featureblobs.append(feature.blob_log(slice, 3**x, 3**x))
# init list of blob coords
blob_coords = []
#print featureblobs
# start at biggest blob size
for featureblob in reversed(featureblobs):
# for every blob found of a blobsize
for blob in enumerate(featureblob):
# if that blob is not within range of another blob, add it
blob = blob[1]
if not within_range(blob, blob_coords):
blob_coords.append([z, blob[0], blob[1], blob[2]])
list.append(blob_coords)
return list
def doh(self):
"""Determinant of Hessian."""
# blob_doh requires double.
image = np.asfarray(self.image, dtype=np.double)
# skimage.feature.blob_doh(image, min_sigma=1, max_sigma=30,
# num_sigma=10, threshold=0.01, overlap=0.5, log_scale=False)
blobs = feature.blob_doh(image, **self.blob_ka, **self.doh_ka)
# Here, radius is approx. equal to sigma.
return blobs
def _identify_features(self, array):
array = (array - array.min()) / array.max()
for time in array.time.values:
array2D = array.sel(time=time).values
array2D.shape()
blobs_doh = blob_doh(array2D, max_sigma=30, threshold=.01)
print(';a')
pass
def process_image(self, new_image):
# Find blobs in the image with scikit-image
self.blobs = blob_doh(new_image,
max_sigma=512,
min_sigma=64,
threshold=.02)
# Send the original image data to the image widget
return new_image
def find_blobs(img_path):
"""Find the blobs in an image"""
from skimage.color import rgb2gray
from skimage.feature import blob_doh
from skimage.io import imread
image = imread(img_path)
image_gray = rgb2gray(image)
return image, blob_doh(image_gray, max_sigma=30, threshold=.01)
def find_blobs(img_path):
"""Find the blobs in an image"""
from skimage.color import rgb2gray
from skimage.feature import blob_doh
from skimage.io import imread
image = imread(img_path)
image_gray = rgb2gray(image)
return image, blob_doh(image_gray, max_sigma=30, threshold=.01)
def get_num_blobs(image):
blobs_doh = blob_doh(image, min_sigma=1, max_sigma=20, threshold=.005)
count = 0
for blob in blobs_doh:
y, x, r = blob
if (x-400)**2 + (y-400)**2 > distance:
continue
count = count + 1
return count
def get_convex_hull_of_blobs_doh(im):
"""
Uses the Difference of Hessian blob detection algorithm from scikit-image to identify the laser points and returns
a ConvexHull of the points found.
:param im: Image with laser points in
:return: ConvexHull of points
"""
# Get blobs using Difference of Hessian
blobs_doh = blob_doh(im, max_sigma=10, num_sigma=1, threshold=.005)
# Compute radii in the 3rd column.
blobs_doh[:, 2] = blobs_doh[:, 2] * sqrt(2) # ToDo: Discount blobs with large radii
if 0 <= len(blobs_doh) < 4:
blobs_doh = blob_doh(im, max_sigma=10, num_sigma=1, threshold=.0025)
elif len(blobs_doh) > 4:
blobs_doh = blob_doh(im, max_sigma=10, num_sigma=1, threshold=.0075)
assert len(blobs_doh) == 4, len(blobs_doh)
return ConvexHull([(blob[0], blob[1]) for blob in blobs_doh])
def evaluateDOH(grayImage):
log = blob_doh(grayImage, max_sigma=30, threshold=.01)
#log[:, 2] = log[:, 2] * sqrt(2)
coordinates = np.empty((log.shape[0], 5))
coordinates[:, 4] = 1
coordinates[:, 0] = log[:, 1] - log[:, 2]
coordinates[:, 1] = log[:, 0] - log[:, 2]
coordinates[:, 2] = log[:, 1] + log[:, 2]
coordinates[:, 3] = log[:, 0] + log[:, 2]
return coordinates
def review_methods(self, image: np.array = None):
"""
Draws a comparison picture of all three methods.
Doesn't work properly with binarized image.
:param image: np.array, str
Image array or image path
:return:
"""
if isinstance(image, str):
image = cv2.imread(image, cv2.COLOR_RGB2GRAY)[..., ::-1]
image_gray = rgb2gray(image)
blobs_log = blob_log(image_gray,
max_sigma=50,
num_sigma=10,
threshold=0.1)
# Compute radii in the 3rd column.
blobs_log[:, 2] = blobs_log[:, 2] * np.sqrt(2)
blobs_dog = blob_dog(image_gray, max_sigma=30, threshold=0.1)
blobs_dog[:, 2] = blobs_dog[:, 2] * np.sqrt(2)
blobs_doh = blob_doh(image_gray, max_sigma=30, threshold=0.01)
blobs_list = [blobs_log, blobs_dog, blobs_doh]
colors = ["yellow", "lime", "red"]
titles = [
"Laplacian of Gaussian",
"Difference of Gaussian",
"Determinant of Hessian",
]
sequence = zip(blobs_list, colors, titles)
fig, axes = plt.subplots(1,
3,
figsize=(9, 3),
sharex=True,
sharey=True)
ax = axes.ravel()
for idx, (blobs, color, title) in enumerate(sequence):
ax[idx].set_title(title)
ax[idx].imshow(image)
for blob in blobs:
y, x, r = blob
c = plt.Circle((x, y), r, color=color, linewidth=2, fill=False)
ax[idx].add_patch(c)
ax[idx].set_axis_off()
plt.tight_layout()
plt.show()
def get_blobs(self, image): blobs_log = blob_log(image, max_sigma=30, num_sigma=10, threshold=.1) blobs_log[:, 2] = blobs_log[:, 2] * sqrt(2) blobs_dog = blob_dog(image, max_sigma=30, threshold=.1) blobs_dog[:, 2] = blobs_dog[:, 2] * sqrt(2) blobs_doh = blob_doh(image, max_sigma=30, threshold=.01) all_blobs = np.vstack([blobs_log, blobs_doh, blobs_dog]) all_blobs = filter(lambda b: b[2] > 4, all_blobs) all_blobs = list(filter(lambda b: b[2] < 60, all_blobs)) return all_blobs
def preview(self, ips, para):
grayimg = ips.img if ips.img.ndim == 2 else ips.img.mean(axis=-1)
grayimg /= grayimg.max()
pts = blob_doh(grayimg,
min_sigma=para['min_sigma'],
max_sigma=para['max_sigma'],
num_sigma=para['num_sigma'],
threshold=para['threshold'],
overlap=para['overlap'],
log_scale=para['log_scale'])
ips.mark = mark2shp({'type': 'circles', 'body': pts[:, [1, 0, 2]]})
def detect_blobs(data_gray):
from skimage.feature import blob_dog, blob_log, blob_doh
from skimage.color import rgb2gray
# takes grayscale
blobs = blob_doh(data_gray, min_sigma=10, num_sigma=3, max_sigma=50, threshold=0.1)
mask = np.zeros(data_gray)
n, m = mask.shape
for blob in blobs:
y, x = np.ogrid[-a : n - a, -b : m - b]
mask = mask * (x * x + y * y <= r * r)
return mask
def find_blobs(img):
img = np.copy(img)
#img[img<0.2*img.max()] = 0
blobs = blob_doh(img, max_sigma=10)
true_blobs = []
for y0, x0, b0 in blobs:
#print img[y0,x0]
#print util.neighbour(img, y0, x0, b0)
#print "---------------------------------"
#print
if img[y0,x0] > util.neighbour(img,y0,x0,b0).mean() and b0 >= 2:
true_blobs.append([y0,x0,b0])
return true_blobs, blobs, img
def test_blob(img, y0, x0, size=40):
""" Test if blob exist at (y0, x0) in image
size: inspecting region
"""
img0 = np.zeros((img.shape[0]+2*size, img.shape[1]+2*size))
img0[size:size+img.shape[0], size:size+img.shape[1]] = img
p0 = util.neighbour(img0, y0+size, x0+size, size)
blobs = blob_doh(p0, max_sigma=10)
for _y, _x, _b in blobs:
if _b > 8: continue
if (_y-size)**2 + (_x-size)**2 <= 4:
return True
return False
def test_blob_overlap():
img = np.ones((512, 512), dtype=np.uint8)
xs, ys = circle(100, 100, 20)
img[xs, ys] = 255
xs, ys = circle(120, 100, 30)
img[xs, ys] = 255
blobs = blob_doh(
img,
min_sigma=1,
max_sigma=60,
num_sigma=10,
threshold=.05)
assert len(blobs) == 1
def blob_feature(image, method='log'):
if method == 'log':
blobs = blob_log(image, )
else:
blobs = blob_doh(image, )
blob_image = np.zeros(image.shape)
#Draw the blobs to an image
for blob in blobs:
y,x,sigma = blob
color = sigma
size = int(np.sqrt(2*sigma)) if method == 'log' else sigma
cv2.circle(blob_image, (x, y), size, sigma/1,-1)
#dataset.show_image(blob_image)
return blob_image
def fun2(fv):
blobs = [ blob_doh(_f) for _f in fv ]
blobs = [ item for sublist in blobs for item in sublist ]
blobs = array(blobs)
if blobs.size == 0:
return 0, 0, 0, False, []
# find maxima
r0, y0, x0 = hough.find_ring(fv)
_d = blobs[:,:2] - [y0, x0]
_d = [ x[0]**2 + x[1]**2 for x in _d ]
_d = [ i for i in _d if i <= 2 ]
if len(_d) >= 1:
check = True
print y0, x0, r0, 'Yes!'
else:
check = False
print y0, x0, r0, 'No!'
#fv[r0, y0-1:y0+2, x0-1:x0+2] = 0
return r0, y0, x0, check, blobs
def detect_blobs(self, min_sigma, max_sigma, num_sigma, threshold):
blobs = []
image_gray = rgb2gray(self.original_image)
r, c = np.shape(image_gray)
image_gray = image_gray[5:r-5,5:c-5]
image_gray[1,1] = 1
if self.method == 'log':
blobs = blob_log(image_gray, min_sigma=min_sigma,
max_sigma=max_sigma, num_sigma=num_sigma, threshold=threshold)
a = len(blobs[:])
if a != 0:
blobs[:, 2] = blobs[:, 2] * sqrt(2)
elif self.method == 'dog':
blobs = blobs_dog = blob_dog(image_gray, max_sigma=max_sigma, threshold=threshold)
a = len(blobs[:])
if a != 0:
blobs[:, 2] = blobs[:, 2] * sqrt(2)
else:
blobs = blob_doh(image_gray, max_sigma=max_sigma, threshold=threshold)
return blobs
def find_blobs_hessian(img):
size = 20
img0 = np.zeros((img.shape[0]+ 2*size, img.shape[1]+ 2*size))
img0[size:size+img.shape[0], size:size+img.shape[1]] = img
blobs = blob_doh(img0, max_sigma=10, num_sigma=10)
true_blobs = []
for y0, x0, b0 in blobs:
p0 = util.neighbour(img0, y0, x0, 1)
y, x = np.unravel_index(p0.argmax(), p0.shape)
y0 += y - 1
x0 += x - 1
p0 = util.neighbour(img0, y0, x0, 2*b0)
if True:
# method 1 (hessian diagonal)
hs0, hs1, hs2 = hessian_matrix(p0, 0.5*b0)
q0 = 0.5 *(hs0 + hs2)
if b0 >= 2 \
and b0 <= 8 \
and q0[2*b0,2*b0] > 0.95*util.neighbour(q0,2*b0,2*b0,b0).max():
true_blobs.append([y0,x0,b0])
else:
# method 2 (hessian determinant)
q0 = hessian_matrix_det(p0, b0)
print b0, q0[2*b0, 2*b0], util.neighbour(q0, 2*b0,2*b0,b0).max()
if b0 >= 2 \
and b0 <= 8 \
and q0[2*b0,2*b0] > 0.8*util.neighbour(q0,2*b0,2*b0,b0).max():
true_blobs.append([y0, x0, b0])
if len(true_blobs) > 0:
true_blobs = np.array(true_blobs) - [size, size, 0]
if len(blobs) > 0:
blobs = blobs - [size, size, 0]
return true_blobs, blobs
def test_blob_overlap():
img = np.ones((256, 256), dtype=np.uint8)
xs, ys = circle(100, 100, 20)
img[xs, ys] = 255
xs, ys = circle(120, 100, 30)
img[xs, ys] = 255
blobs = blob_doh(
img,
min_sigma=1,
max_sigma=60,
num_sigma=10,
threshold=.05)
assert len(blobs) == 1
r1, r2 = 7, 6
pad1, pad2 = 11, 12
blob1 = ellipsoid(r1, r1, r1)
blob1 = util.pad(blob1, pad1, mode='constant')
blob2 = ellipsoid(r2, r2, r2)
blob2 = util.pad(blob2, [(pad2, pad2), (pad2 - 9, pad2 + 9),
(pad2, pad2)],
mode='constant')
im3 = np.logical_or(blob1, blob2)
blobs = blob_log(im3, min_sigma=2, max_sigma=10, overlap=0.1)
assert len(blobs) == 1
# Two circles with distance between centers equal to radius
overlap = _blob_overlap(np.array([0, 0, 10 / math.sqrt(2)]),
np.array([0, 10, 10 / math.sqrt(2)]))
assert_almost_equal(overlap,
1./math.pi * (2 * math.acos(1./2) - math.sqrt(3)/2.))
def blob_image(image):
#img_path = '../1.3.6.1.4.1.14519.5.2.1.6279.6001.100332161840553388986847034053.mhd'
# slice = numpy_image[240,:,:]
# normalized = norm.normalize(return_surrounding([240,240,240],numpy_image, 240))
# thresholded = threshold_by_histogram(normalized)
# blobs = label_image(thresholded)
# show_images([blobs, normalized, thresholded])
# normalized3d = norm.normalize(numpy_image)
# thresholded3d = threshold_by_histogram(normalized3d)
list = []
image = norm.normalize(image)
#print "normalized and thresholded"
for z, slice in tqdm(enumerate(image)):
blobs = feature.blob_doh(slice)
#print blobs.shape
blob_coords = np.zeros((len(blobs),3))
for i, blob in enumerate(blobs):
blob_coords[i] = [z, blob[0], blob[1]]
list.append(blob_coords)
#print list
return list
from skimage.feature import blob_dog, blob_log, blob_doh
from math import sqrt
from skimage.color import rgb2gray
# image = data.hubble_deep_field()[0:500, 0:500]
image = cv2.resize(img1, (256, 256))
image_gray = rgb2gray(image)
blobs_log = blob_log(image_gray, max_sigma=30, num_sigma=10, threshold=0.1)
# Compute radii in the 3rd column.
blobs_log[:, 2] = blobs_log[:, 2] * sqrt(2)
blobs_dog = blob_dog(image_gray, max_sigma=30, threshold=0.1)
blobs_dog[:, 2] = blobs_dog[:, 2] * sqrt(2)
blobs_doh = blob_doh(image_gray, max_sigma=30, threshold=0.01)
blobs_list = [blobs_log, blobs_dog, blobs_doh]
colors = ["yellow", "lime", "red"]
titles = ["Laplacian of Gaussian", "Difference of Gaussian", "Determinant of Hessian"]
sequence = zip(blobs_list, colors, titles)
for blobs, color, title in sequence:
fig, ax = plt.subplots(1, 1)
ax.set_title(title)
ax.imshow(image, interpolation="nearest")
for blob in blobs:
y, x, r = blob
c = plt.Circle((x, y), r, color=color, linewidth=2, fill=False)
ax.add_patch(c)
test_images = [lensed_image, unlensed_image]
gray_images = [rgb2gray(image) for image in test_images]
blobs_log = [blob_log(image_gray, max_sigma=30, num_sigma=10, threshold=.1) for image_gray in gray_images]
# Compute radii in the 3rd column.
# for bl in blobs_log :
# bl[:, 2] = bl[:, 2] * sqrt(2)
# blobs_dog = [blob_dog(image_gray, max_sigma=30, threshold=.1) for image_gray in gray_images]
# for bd in blobs_dog :
# bd[:, 2] = bd[:, 2] * sqrt(2)
blobs_doh = [blob_doh(image_gray, max_sigma=30, threshold=.01) for image_gray in gray_images]
print('Finished blob detection with Determinant of Hessian')
blobs_list = [#blobs_log, blobs_dog,
blobs_doh]
colors = ['yellow', 'lime', 'red']
titles = ['Laplacian of Gaussian', 'Difference of Gaussian',
'Determinant of Hessian']
sequence = zip(blobs_list, colors, titles)
for i, image in enumerate(['lensed','unlensed']) :
fig, axes = plt.subplots(1, 3, figsize=(14, 4), sharex=True, sharey=True,
subplot_kw={'adjustable': 'box-forced'})
plt.tight_layout()
def create_blobs(url, opt='all'):
image = imread(url)
return_dict = dict()
print 'Image read: {}'.format(url)
image_gray = rgb2gray(image)
print 'Greyscale applied'
if opt=='all':
blobs_log = blob_log(image_gray, min_sigma=15, max_sigma=50, num_sigma=10, threshold=.1, overlap=0.8)
print 'Laplacian of Gaussian computed'
# Compute radii in the 3rd column.
blobs_log[:, 2] = blobs_log[:, 2] * sqrt(2)
return_dict['LoG'] = blobs_log
blobs_dog = blob_dog(image_gray, max_sigma=30, threshold=.1)
print 'Difference of Gaussian computed'
blobs_dog[:, 2] = blobs_dog[:, 2] * sqrt(2)
return_dict['DoG'] = blobs_dog
blobs_doh = blob_doh(image_gray, max_sigma=30, threshold=.01)
print 'Determinant of Hessian computed'
return_dict['DoH'] = blobs_doh
blobs_list = [blobs_log, blobs_dog, blobs_doh]
colors = ['yellow', 'lime', 'red']
titles = ['LoG', 'DoG',
'DoH']
sequence = zip(blobs_list, colors, titles)
print 'Sequence created'
elif opt=='doh':
print 'DoH only'
blobs_doh = blob_doh(image_gray, max_sigma=30, threshold=.01)
return_dict['DoH'] = blobs_doh
sequence = zip([blobs_doh], ['red'], ['DoH'])
elif opt=='log':
print 'LoG only'
blobs_log = blob_log(image_gray, min_sigma=15, max_sigma=50, num_sigma=10, threshold=.1, overlap=0.8)
print 'Laplacian of Gaussian computed'
# Compute radii in the 3rd column.
blobs_log[:, 2] = blobs_log[:, 2] * sqrt(2)
return_dict['LoG'] = blobs_log
sequence = zip([blobs_log], ['yellow'], ['LoG'])
print sequence
fig,axes = plt.subplots(1, 3, sharex=True, sharey=True, subplot_kw={'adjustable':'box-forced'})
axes = axes.ravel()
print 'Matplot initialized'
print sequence
for blobs, color, title in sequence:
ax = axes[0]
axes = axes[1:]
ax.set_title(title)
ax.imshow(image, interpolation='nearest')
for blob in blobs:
y, x, r = blob
c = plt.Circle((x, y), r, color=color, linewidth=2, fill=False)
ax.add_patch(c)
plt.show()
return return_dict
#blobs = blob_doh(image_gray, min_sigma=1, max_sigma=25, num_sigma=15, threshold=.001)
blobs = blob_dog(image_gray, min_sigma=1, max_sigma=25, sigma_ratio=1.6, threshold=.25, overlap=0.5)
if(len(blobs) != 0):
blobs[:, 2] = blobs[:, 2] * sqrt(2)
self.all_blobs.append(blobs)
file_processing = "Processing files: " + str(i+1) + "/" + str(len(self.filenames))
#print("blobs size: ", len(blobs))
=======
#image = cv2.imread(glob.glob("cropped_images/" + self.filenames[i]))
#print("imageName: ", (glob.glob("cropped_images/" + self.filenames[i]))[0])
image = cv2.imread((glob.glob("cropped_images/*" + self.filenames[i]))[0])
self.images.append(image)
image_gray = rgb2gray(image)
#blobs = blob_doh(image_gray, min_sigma=3, max_sigma=35, num_sigma=30, threshold=.005)
blobs = blob_doh(image_gray, min_sigma=1, max_sigma=25, num_sigma=15, threshold=.001)
self.all_blobs.append(blobs)
file_processing = "Processing files: " + str(i+1) + "/" + str(len(self.filenames))
print("blobs size: ", len(blobs))
>>>>>>> c1b308ec15ebe3098d38e2f3bd0171bbd669606b
print(file_processing)
def sortKey(self, str):
i = str.find("frame")
j = str.find(".jpg")
<<<<<<< HEAD
=======
#print("shortened string: ", str[i+5:j])
>>>>>>> c1b308ec15ebe3098d38e2f3bd0171bbd669606b
f1 = copy(f0[0, 100:1100, 50:1050])
f1 = img_as_float(f1)
#~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
# Settings
#~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
bmin = 0.12
bmax = 0.20
rmin = 2
rmax = 65
alpha = 0.5
# detek very bright blob and remove them
blobs = blob_doh(f1, threshold=0.002)
for b in blobs:
b[2] *= 1.2
f1[b[0]-b[2]:b[0]+b[2]+1, b[1]-b[2]:b[1]+b[2]+1] = f1.mean()
# downsize
# threholding filter
# sample importance
f2 = rescale(f1, 0.5)
f3 = ifilter.threhold(f2, bmin, bmax)
f4 = ifilter.sample_importance(f3, alpha)
# hough transform
if False:
filename = 'C:\\Users\\schan3\\Desktop\\New folder (2)\\'
imgname = 'SBU_2015.01.26_18.21.27_flake_11032_cam_1'
file_processed = 0
for fname in glob.glob(filename + '*.png'):
count = 0
print fname
image = io.imread(fname)
image_gray = rgb2gray(image)
blobs_doh = blob_doh(image_gray, max_sigma=1000, min_sigma =40, threshold=.00045, overlap = .25)
blobs_list = [blobs_doh]
colors = ['red']
titles = ['Snow flake counts using Determinant of Hessian']
sequence = zip(blobs_list, colors, titles)
for blobs, color, title in sequence:
fig, ax = plt.subplots(1, 1)
ax.set_title(title)
ax.imshow(image, interpolation='nearest', cmap = cm.Greys_r)
for blob in blobs:
y, x, r = blob
c = plt.Circle((x, y), r, color=color, linewidth=1, fill=False)
ax.add_patch(c)
if flip_v:
cv2.flip(a,0,a)
cv2.normalize(a, a, 0, 65535, cv2.NORM_MINMAX)
np.right_shift(a, 8, a)
return np.uint8(a)
camera = picamera.PiCamera()
url = "http://52.90.77.195"
# get video from flir or infinite loop
while True:
cv2.imwrite("wat.jpg", capture())
image = imread("wat.jpg") # gonna be frame
image_gray = rgb2gray(image) # convert frame
blobs_doh = blob_doh(image_gray, min_sigma=20, max_sigma=35, threshold=.01) # detect frame
if len(blobs_doh):
file = ''.join(random.choice(string.ascii_uppercase + string.digits) for _ in range(10)) + '.jpg'
camera.capture(file)
print "request to server here to drone"
r = requests.get(url + "/intrude")
if(start < 70):
start = start + 15
pwm.ChangeDutyCycle(start)
else:
start = 5
pwm.ChangeDutyCycle(start)
time.sleep(0.5)
# potentially move to blob
y = blobs_doh[0][0]
def test_blob_doh():
img = np.ones((512, 512), dtype=np.uint8)
img3 = np.ones((5, 5, 5))
xs, ys = circle(400, 130, 20)
img[xs, ys] = 255
xs, ys = circle(460, 50, 30)
img[xs, ys] = 255
xs, ys = circle(100, 300, 40)
img[xs, ys] = 255
xs, ys = circle(200, 350, 50)
img[xs, ys] = 255
blobs = blob_doh(
img,
min_sigma=1,
max_sigma=60,
num_sigma=10,
threshold=.05)
radius = lambda x: x[2]
s = sorted(blobs, key=radius)
thresh = 4
b = s[0]
assert abs(b[0] - 400) <= thresh
assert abs(b[1] - 130) <= thresh
assert abs(radius(b) - 20) <= thresh
b = s[1]
assert abs(b[0] - 460) <= thresh
assert abs(b[1] - 50) <= thresh
assert abs(radius(b) - 30) <= thresh
b = s[2]
assert abs(b[0] - 100) <= thresh
assert abs(b[1] - 300) <= thresh
assert abs(radius(b) - 40) <= thresh
b = s[3]
assert abs(b[0] - 200) <= thresh
assert abs(b[1] - 350) <= thresh
assert abs(radius(b) - 50) <= thresh
# Testing log scale
blobs = blob_doh(
img,
min_sigma=1,
max_sigma=60,
num_sigma=10,
log_scale=True,
threshold=.05)
b = s[0]
assert abs(b[0] - 400) <= thresh
assert abs(b[1] - 130) <= thresh
assert abs(radius(b) - 20) <= thresh
b = s[1]
assert abs(b[0] - 460) <= thresh
assert abs(b[1] - 50) <= thresh
assert abs(radius(b) - 30) <= thresh
b = s[2]
assert abs(b[0] - 100) <= thresh
assert abs(b[1] - 300) <= thresh
assert abs(radius(b) - 40) <= thresh
b = s[3]
assert abs(b[0] - 200) <= thresh
assert abs(b[1] - 350) <= thresh
assert abs(radius(b) - 50) <= thresh
with pytest.raises(ValueError):
blob_doh(img3)
# Testing no peaks
img_empty = np.zeros((100,100))
assert blob_doh(img_empty).size == 0
plt.imshow(image_gray, cmap='gray')
plt.show()
# blobs_log = feature.blob_log(image_gray, max_sigma=30, num_sigma=10, threshold=.1)
# plt.imshow(blobs_log, cmap='gray')
# plt.show()
blobs_log = feature.blob_log(image_gray, max_sigma=30, num_sigma=10, threshold=.1)
# Compute radii in the 3rd column.
blobs_log[:, 2] = blobs_log[:, 2] * sqrt(2)
blobs_dog = feature.blob_dog(image_gray, max_sigma=30, threshold=.1)
blobs_dog[:, 2] = blobs_dog[:, 2] * sqrt(2)
blobs_doh = feature.blob_doh(image_gray, max_sigma=30, threshold=.01)
blobs_list = [blobs_log, blobs_dog, blobs_doh]
colors = ['yellow', 'lime', 'red']
titles = ['Laplacian of Gaussian', 'Difference of Gaussian', 'Determinant of Hessian']
sequence = zip(blobs_list, colors, titles)
fig, axes = plt.subplots(1, 3, figsize=(14, 4), sharex=True, sharey=True,
subplot_kw={'adjustable': 'box-forced'})
plt.tight_layout()
axes = axes.ravel()
for blobs, color, title in sequence:
ax = axes[0]
axes = axes[1:]
ax.set_title(title)