Apply openCV Look Up Table (LUT) to an image

Question

I want to apply a custom color look up table (LUT) on an image.

I have looked over the openCV LUT Api but I can not seem to get it right.

#!/bin/bash
# -*- coding: utf-8 -*-

import os
import numpy as np
import cv2

b_max = 230
g_max = 220
r_max = 250

identity = np.arange(256, dtype = np.dtype('uint8'))

b_channel = np.arange(b_max, dtype = np.dtype('uint8'))
b_channel = np.append(b_channel, (256-b_max)*[b_max])
g_channel = np.arange(g_max, dtype = np.dtype('uint8'))
g_channel = np.append(g_channel, (256-g_max)*[g_max])
r_channel = np.arange(r_max, dtype = np.dtype('uint8'))
r_channel = np.append(r_channel, (256-r_max)*[r_max])

if 256 != b_channel.size or 256 != g_channel.size or 256 != r_channel.size:
    print "size of arrays don't match!"

lut = np.dstack((identity, identity, b_channel, g_channel, r_channel))

# Load the image
img = cv2.imread('input.jpg',cv2.IMREAD_COLOR)
dstImage = cv2.LUT(img, lut)
cv2.imwrite('output.jpg', dstImage)

I get the following error:

OpenCV Error: Assertion failed ((lutcn == cn || lutcn == 1) && _lut.total() == 256 && _lut.isContinuous() && (depth == CV_8U || depth == CV_8S)) in cv::LUT, file C:\build\master_winpack-bindings-win32-vc14-static\opencv\modules\core\src\convert.cpp, line 5690
Traceback (most recent call last):
  File "ApplyCLUT.py", line 33, in <module>
    dstImage = cv2.LUT(img, lut)
cv2.error: C:\build\master_winpack-bindings-win32-vc14-static\opencv\modules\core\src\convert.cpp:5690: error: (-215) (lutcn == cn || lutcn == 1) && _lut.total() == 256 && _lut.isContinuous() && (depth == CV_8U || depth == CV_8S) in function cv::LUT

EDIT: change to this line:

lut = np.dstack( (b_channel, g_channel, r_channel) )

and it will work!

Answer 1

Below is an example of building LUT:

identity = np.arange(256, dtype = np.dtype('uint8'))
zeros = np.zeros(256, np.dtype('uint8'))
lut = np.dstack((identity, identity, zeros))

Note, there is much simpler and straightforward way zeroing the red channel:

img[:,:,2] = 0

Also calling LUT function should look like this:

dstImage = cv2.LUT(img, lut)

Set blue values greater than b_max to b_max:

b_max = 20
img[img[:,:,0] > b_max, 0] = b_max;