haar_like_feature generates Nan's

[MrAnisotropy]

haar_like_feature generates Nan's

I am using haar_like_feature in skimage and I am getting nan values for certain features. I have tested using various shape sizes, transposing the image, normalizing, making the dimensions odd and even but the problem remains.

For the minimal test below with a small test data of 17x9 I get 123 nan's. Checking the documentation and codes suggests that there is not a limitation of array shapes and I think the inputs are all correct.

If you can take a look I would be grateful. Thanks

Minimal Example

import numpy as np
import skimage.transform,skimage.feature

np.random.seed(0)
image         = np.random.random( (17,9) )

integralImage = skimage.transform.integral_image(image)

test          = skimage.feature.haar_like_feature(integralImage,0,0,\
                                                  integralImage.shape[0],integralImage.shape[1],\
                                                  feature_type=['type-3-x','type-3-y'])
print (np.sum(np.isnan(test)))

Version information

This was tested on Windows 7 and 10.

# Paste the output of the following python commands
from __future__ import print_function
import sys; print(sys.version)
import platform; print(platform.platform())
import skimage; print("scikit-image version: {}".format(skimage.__version__))
import numpy; print("numpy version: {}".format(numpy.__version__))

3.8.5 (default, Sep  3 2020, 21:29:08) [MSC v.1916 64 bit (AMD64)]

Windows-10-10.0.18362-SP0

scikit-image version: 0.17.2

numpy version: 1.19.2

[MrAnisotropy]

Turns out I had the width and height parameters flipped ! Closing the issue.

[MrAnisotropy]

Flipped parameters for width and height. Closing the issue. Thanks