Revision history [back]

CalcHist gives different results on C++ vs Python

I'm loading a Tiff and running calcHist in python and c++, but I can't get the results to match. Code and output attached.

Thank you!

Python Implementation:

import numpy as np
import cv2 as cv
from PIL import Image
import matplotlib.pyplot as plt
im = Image.open('data.tif')
img = np.array(im).astype(np.uint16)
hist = cv.calcHist([img],[0],None,[256],[0,334])
print(hist)
plt.hist(img.ravel(),255,range=(0,1305)); plt.show()

C++:

uint16* raster = new uint16[width*length];
for (uint16 y = 0; y < length; ++y) {
    for (uint16 x = 0; x < width; ++x) {
        raster[y*width + x] = rasterYX[y][x];
    }
}

cv::Mat flattened(length, width, CV_16UC1, &raster[0]);
cv::Mat inputHist;

cv::Mat hist;
float range[] = {0,334};
int num_bins = 256;
const float* ranges[] = {range};
cv::calcHist(&flattened, 1, 0, cv::Mat(), hist, 1,&num_bins, ranges, true, false);

std::cout << "Hist  = " << std::endl << cv::format(hist, cv::Formatter::FMT_PYTHON);

Here is part of the c++ output followed by the python output at the point where they begin to diverge.

0, 1, 2, 1, 4, 4, 2, 1, 11, 6, 16, 38,

[0.000000e+00] [1.000000e+00] [2.000000e+00] [1.000000e+00] [5.000000e+00] [3.000000e+00] [2.000000e+00] [6.000000e+00] [6.000000e+00] [6.000000e+00] [3.100000e+01] [2.300000e+01] [3.600000e+01] [2.010000e+02] [2.160000e+02] [4.350000e+02] [9.560000e+02]