給定兩個或更多 colors，如何找到最大和最小顏色值？

Question

I'm processing images using opencv and part of my problem requires that I iterate over the image by its colors and only display colors which match certain ranges of colors. 如何找到最大和最小 bgr 值？

Answer 1

這是在 Python/OpenCV/Numpy 中執行此操作的一種方法。 你想要的顏色的閾值。 然后使用 Numpy 得到每個通道的最小和最大 colors。

輸入：

import cv2
import numpy as np

# load image 
img = cv2.imread('mandril3.jpg')

# create mask for red
lower=np.array((20,40,215))
upper=np.array((100,120,255))
mask = cv2.inRange(img, lower, upper)

# mask the image for viewing
result = img.copy()
result[mask!=255] = (0,0,0)

# separate channels
b,g,r = cv2.split(img)

# compute min and max ranges of red
bmin = np.amin(b[np.where(mask == 255)])
bmax = np.amax(b[np.where(mask == 255)])
gmin = np.amin(g[np.where(mask == 255)])
gmax = np.amax(g[np.where(mask == 255)])
rmin = np.amin(r[np.where(mask == 255)])
rmax = np.amax(r[np.where(mask == 255)])
print("min red:", bmin, gmin, rmin)
print("max red:", bmax, gmax, rmax)

# save results
cv2.imwrite("mandril3_nose_mask.jpg", mask)
cv2.imwrite("mandril3_nose.jpg", result)

cv2.imshow("mask", mask)
cv2.imshow("result", result)
cv2.waitKey(0)
cv2.destroyAllWindows()

紅鼻子的最小和最大 Colors：

最小紅色：20 53 215 最大紅色：100 120 255

面具：

蒙面圖像：

Answer 2

這是在 Python/OpenCV 中執行此操作的更合適的方法。 我們基本上找到了 R,G,B 通道在屏蔽紅色后的平均值最小的顏色，而 R,G,B 的平均值在屏蔽紅色后的顏色最大。

 - Read the input
 - Select color -- in this case red
 - Threshold the image for the color as a mask
 - Separate channels
 - Get the average image of the 3 channels
 - Find the min and max values in the average image
 - Create a mask for the min values
 - Create a mask for the max values
 - Combine the original mask with the min mask
 - combine the original mask with the max mask
 - Get the coordinates for all pixels where the combined mask and min mask is white.
 - Find the color corresponding to the first coordinate found if more than one
 - Get the coordinates for all pixels where the combined mask and max mask is white.
 - Find the color corresponding to the first coordinate found if more than one

輸入：

import cv2
import numpy as np

# load image 
img = cv2.imread('mandril3.jpg')

# create mask for red
lower=np.array((20,40,215))
upper=np.array((100,120,255))
mask = cv2.inRange(img, lower, upper)

# mask the image for viewing
result = img.copy()
result[mask!=255] = (0,0,0)

# separate channels
b,g,r = cv2.split(img)

# get average of channels
ave = cv2.add(b,g,r)/3
ave = ave.astype(np.uint8)

# get min and max for ave
ave_min = np.amin(ave[np.where(mask==255)])
ave_max = np.amax(ave[np.where(mask==255)])

# form min and max masks from ave
mask_min = ave.copy()
mask_min[ave==ave_min] = 255
mask_min[ave!=ave_min] = 0

mask_max = ave.copy()
mask_max[ave==ave_max] = 255
mask_max[ave!=ave_max] = 0

# combine min mask with red mask
red_mask_min = cv2.bitwise_and(mask, mask_min)

# combine max mask with red mask
red_mask_max = cv2.bitwise_and(mask, mask_max)

# get coordinates where masks are white and corresponding color - take first one
min_list = np.argwhere(red_mask_min==255)
min_count = len(min_list)
if min_count != 0:
    min_coord = min_list[0]
    y=min_coord[0]
    x=min_coord[1]
    min_color = img[y:y+1, x:x+1][0][0]
    print("min_red:", min_color)

max_list = np.argwhere(red_mask_max==255)
max_count = len(max_list)
if max_count != 0:
    max_coord = max_list[0]
    y=max_coord[0]
    x=max_coord[1]
    max_color = img[y:y+1, x:x+1][0][0]
    print("max_red:", max_color)


# save results
cv2.imwrite("mandril3_nose_mask.jpg", mask)
cv2.imwrite("mandril3_nose.jpg", result)

# show images
cv2.imshow("mask", mask)
cv2.imshow("result", result)
cv2.imshow("ave", ave)
cv2.imshow("red_mask_min", red_mask_min)
cv2.imshow("red_mask_max", red_mask_max)
cv2.waitKey(0)
cv2.destroyAllWindows()

紅色面具：

蒙面圖像：

結果：

min_red: [ 22  57 244]
max_red: [ 99 120 242]