How to calculate 3D histogram in python using open CV

Question

I want to calculate 3D histogram of my Cielab image in python. I am using openCV to calculate my histogram. I want to compare images using compareHist function

Answer 1

For computing a histogram in cielab, this worked for me:

function signature of calcHist: images, channels, mask, number of bins, array of the dims arrays of the histogram bin boundaries in each dimension.

img = cv2.imread(file) # in bgr 
hist = cv2.calcHist([img],[0, 1, 2],None,[256, 256, 256],[0, 255, 0, 255, 0, 255]) # in bgr
img = cv2.cvtColor(img, cv2.COLOR_BGR2LAB) # in lab 
hist = cv2.calcHist([img],[0, 1, 2],None,[100, 2*128, 2*128],[0, 100, -128, 127, -128, 127])`# in lab 

correlation = cv2.HISTCMP_CORREL # compare histograms using correlation
corr = cv2.compareHist(img, img2, correlation)

Answer 2

I came across this question while trying to make a 3D histogram of an HSV image, and encountered the same error. It turns out that the OpenCV documentation is leading us astray here. The docs are written for the C++ API and as such can only be used as a vague guide to the Python cv2 API (although I have found that the docs are misleading for C++ as well at times).

The function signature is as follows:

cv2.calcHist(images, channels, mask, histSize, ranges[, hist[, accumulate]]) -> hist

The key point is that the channels, histSize and ranges parameters should be flat lists, not nested lists as in your example. Try the following, assuming i_lab is a three-channel image:

range_hist = [0, 100, -100, 100, -100, 100]
hist_1 = cv2.calcHist([i_lab], [0, 1, 2], None, [20, 20, 20], range_hist)

For a more complete example, try this code listing from the opencvpython blog.