def delta_e_cie2000(color1, color2, Kl=1, Kc=1, Kh=1):
"""
Calculates the Delta E (CIE2000) of two colors.
"""
color1_vector = _get_lab_color1_vector(color1)
color2_matrix = _get_lab_color2_matrix(color2)
delta_e = color_diff_matrix.delta_e_cie2000(
color1_vector, color2_matrix, Kl=Kl, Kc=Kc, Kh=Kh)[0]
return numpy.asscalar(delta_e)
def delta_e_matrix(self, lab_color_matrix, mode='cie2000', *args, **kwargs):
"""
Compares this object with all colors in lab_color_matrix via Delta E.
The matrix must be of shape (n,3) and must be composed of floats.
Returns a distance vector of shape (n,).
Valid modes:
cie2000
cie1976
cie1994
cmc
"""
lab_color_vector = np.array([self.lab_l, self.lab_a, self.lab_b])
mode = mode.lower()
if mode == 'cie2000':
return color_diff_matrix.delta_e_cie2000(lab_color_vector, lab_color_matrix)
elif mode == 'cie1994':
return color_diff_matrix.delta_e_cie1994(lab_color_vector, lab_color_matrix, **kwargs)
elif mode == 'cie1976':
return color_diff_matrix.delta_e_cie1976(lab_color_vector, lab_color_matrix)
elif mode == 'cmc':
return color_diff_matrix.delta_e_cmc(lab_color_vector, lab_color_matrix, **kwargs)
else:
raise InvalidDeltaEMode(mode)
def labColor2LystColor(target_color):
# target_color is a lab converted color
lab_color = getLystLabColor()
lab_matrix = getLystColorMatrix()
lab_color_vector = np.array(
[target_color.lab_l, target_color.lab_a, target_color.lab_b])
delta = delta_e_cie2000(lab_color_vector, lab_matrix)
color = lab_color[np.argmin(delta)]
return color
def is_close(x0, y0, x1, y1):
global thresh, image
color_vec = np.array(image[x0, y0, :])
color_mat = np.array([(image[x1, y1, 0], image[x1, y1, 1], image[x1, y1,
2])])
dist = np.asscalar(delta_e_cie2000(color_vec, color_mat)[0])
print(dist)
if dist < thresh:
return True
return False
def delta_e_cie2000(color1, color2, Kl=1, Kc=1, Kh=1):
"""
Calculates the Delta E (CIE2000) of two colors.
"""
from warnings import warn
warn("The current form of this class being depreciated in favor of not passing in arguments for Kl, Kc, and Kh when unity 1 is used. For applications wishing to retain the ability to change Kl, Kc, and/or Kh from default value of 1 see delta_e_cie_2000_nonunity. In a future release, the updated format as seen in delta_e_cie2000_update will be used")
color1_vector = _get_lab_color1_vector(color1)
color2_matrix = _get_lab_color2_matrix(color2)
delta_e = color_diff_matrix.delta_e_cie2000(
color1_vector, color2_matrix, Kl=Kl, Kc=Kc, Kh=Kh)[0]
return numpy.asscalar(delta_e)
def get_color_name(rgb_vector):
##Find color - https://making.lyst.com/2014/02/22/color-detection/
# load list of 1000 random colors from the XKCD color chart
# with open('colors.csv', 'rb') as csvfile:
# reader = csv.reader(csvfile)
# rgb_matrix = np.array([map(float, row[0:3]) for row in reader],dtype=float)
# with open('colors.csv', 'rb') as csvfile:
# reader = csv.reader(csvfile)
# color_labels= np.array([row[3:] for row in reader])
# # the reference color
# lab_matrix=np.array([convert_color(sRGBColor(row[0],row[1],row[2],is_upscaled=True),LabColor) for row in rgb_matrix])
# print lab_matrix
# with open('lab_colors.csv', 'wb') as csvfile:
# writer = csv.writer(csvfile, delimiter=',',
# quotechar='|', quoting=csv.QUOTE_MINIMAL)
# for idx,row in enumerate(lab_matrix):
# writer.writerow([row.lab_l,row.lab_a,row.lab_b,color_labels[idx][0]])
with open(
os.path.join(os.path.dirname(os.path.realpath(__file__)),
'lab_colors.csv'), 'rb') as csvfile:
reader = csv.reader(csvfile)
lab_matrix = np.array([map(float, row[0:3]) for row in reader],
dtype=float)
color_index = 0
lab_color = convert_color(
sRGBColor(rgb_vector[0],
rgb_vector[1],
rgb_vector[2],
is_upscaled=True), LabColor)
# writer.writerow([row.lab_l,row.lab_a,row.lab_b,color_labels[idx][0]])
with open(
os.path.join(os.path.dirname(os.path.realpath(__file__)),
'lab_colors.csv'), 'rb') as csvfile:
reader = csv.reader(csvfile)
color_labels = np.array([row[3:] for row in reader])
_color = []
_color.append(lab_color.lab_l)
_color.append(lab_color.lab_a)
_color.append(lab_color.lab_b)
nearest_color = lab_matrix[np.argmin(delta_e_cie2000(_color, lab_matrix))]
color_idx = 0
for idx, item in enumerate(lab_matrix):
if item[0] == nearest_color[0] and item[1] == nearest_color[
1] and item[2] == nearest_color[2]:
color_idx = idx
# Find the color difference
#print '%s OR %s is closest to %s %s' % (rgb_vector,_color, nearest_color,color_labels[color_idx])
return color_labels[color_idx]
def delta_e_cie2000(color1, color2, Kl=1, Kc=1, Kh=1):
"""
Calculates the Delta E (CIE2000) of two colors.
"""
from warnings import warn
warn(
"The current form of this class being depreciated in favor of not passing in arguments for Kl, Kc, and Kh when unity 1 is used. For applications wishing to retain the ability to change Kl, Kc, and/or Kh from default value of 1 see delta_e_cie_2000_nonunity. In a future release, the updated format as seen in delta_e_cie2000_update will be used"
)
color1_vector = _get_lab_color1_vector(color1)
color2_matrix = _get_lab_color2_matrix(color2)
delta_e = color_diff_matrix.delta_e_cie2000(color1_vector,
color2_matrix,
Kl=Kl,
Kc=Kc,
Kh=Kh)[0]
return numpy.asscalar(delta_e)
def extract_colors(img_path, color_lab_matrix_path, color_names_path, labels,
merge_labels):
quality = 1
lab_matrix = pickle.load(open(color_lab_matrix_path, 'rb'))
color_names = pickle.load(open(color_names_path, 'rb'))
img = imread(img_path)
num_rows = img.shape[0]
num_cols = img.shape[1]
num_pixels = num_rows * num_cols
ce = ColorExtractor()
image_attributes = {}
image_attributes['height'] = num_rows
image_attributes['width'] = num_cols
#image_attributes['name'] = im_prefix + ext
category_labels = category_original_labels
img_flat = img.transpose(2, 0, 1).reshape(3, -1).T
if merge_labels == True:
labels = label_merging(labels, category_labels)
category_labels = category_merged_labels
labels_flat = np.ravel(labels)
uniq_labels = np.unique(labels_flat)
color_pallete = {}
for ind in list(uniq_labels):
img_region = img_flat[labels_flat == ind]
pixel_list = list(tuple(map(tuple, img_region)))
num_colors = 5
labels_region = (labels == ind)
ymax, xmax = np.max(np.where(labels_region == True), 1)
ymin, xmin = np.min(np.where(labels_region == True), 1)
region_box = [xmin, ymin, xmax, ymax]
pallete = ce.get_palette(pixel_list,
color_count=num_colors,
quality=quality)
pallete = pallete[0]
pixel_area = round((100.0 * img_region.shape[0]) / num_pixels, 2)
if pixel_area < 1:
continue
r = pallete['r'] / 255.0
g = pallete['g'] / 255.0
b = pallete['b'] / 255.0
color_rgb = sRGBColor(r, g, b)
color_lab = convert_color(color_rgb, LabColor)
color_lab_vec = np.array(
[color_lab.lab_l, color_lab.lab_a, color_lab.lab_b])
delta = delta_e_cie2000(color_lab_vec, lab_matrix)
color_name = color_names[np.argmin(delta)]
rgb_values = {}
rgb_values['r'] = pallete['r']
rgb_values['g'] = pallete['g']
rgb_values['b'] = pallete['b']
color_pallete[category_labels[ind]] = {
'color': color_name,
'pixel_area': pixel_area,
'box': region_box,
'rgb': rgb_values
}
image_attributes['info'] = color_pallete
return image_attributes
cPickled (n,3) numpy array LAB values such that row q maps to
index q in the lab color list
"""
import sys
import csv
import bz2
import numpy as np
# Does some sys.path manipulation so we can run examples in-place.
# noinspection PyUnresolvedReferences
import example_config
from colormath.color_diff_matrix import delta_e_cie2000
from colormath.color_objects import LabColor
# load list of 1000 random colors from the XKCD color chart
if sys.version_info >= (3, 0):
reader = csv.DictReader(bz2.open('lab_matrix.csv.bz2', mode='rt'))
lab_matrix = np.array([list(map(float, row.values())) for row in reader])
else:
reader = csv.DictReader(bz2.BZ2File('lab_matrix.csv.bz2'))
lab_matrix = np.array([map(float, row.values()) for row in reader])
color = LabColor(lab_l=69.34, lab_a=-0.88, lab_b=-52.57)
lab_color_vector = np.array([color.lab_l, color.lab_a, color.lab_b])
delta = delta_e_cie2000(lab_color_vector, lab_matrix)
print('%s is closest to %s' % (color, lab_matrix[np.argmin(delta)]))
index q in the lab color list
"""
import sys
import csv
import bz2
import numpy as np
# Does some sys.path manipulation so we can run examples in-place.
# noinspection PyUnresolvedReferences
import example_config
from colormath.color_diff_matrix import delta_e_cie2000
from colormath.color_objects import LabColor
# load list of 1000 random colors from the XKCD color chart
if sys.version_info >= (3, 0):
reader = csv.DictReader(bz2.open('lab_matrix.csv.bz2', mode='rt'))
lab_matrix = np.array([list(map(float, row.values())) for row in reader])
else:
reader = csv.DictReader(bz2.BZ2File('lab_matrix.csv.bz2'))
lab_matrix = np.array([map(float, row.values()) for row in reader])
color = LabColor(lab_l=69.34, lab_a=-0.88, lab_b=-52.57)
lab_color_vector = np.array([color.lab_l, color.lab_a, color.lab_b])
delta = delta_e_cie2000(lab_color_vector, lab_matrix)
print('%s is closest to %s' % (color, lab_matrix[np.argmin(delta)]))
