Hi!
I'm not sure if this is a bug, or me misunderstanding some part of the API, or an implementation limitation: I get quite different, and for me, worse, results from the OpenCV 2 implementation of the probabalistic Hough lines transform, compared to the Scikit Image one. This is using the exact same parameters for the two implementations.
To be clear, I'm comparing these two functions:
https://docs.opencv.org/2.4/modules/imgproc/doc/feature_detection.html?highlight=houghlinesp#houghlinesp http://scikit-image.org/docs/dev/api/skimage.transform.html#skimage.transform.probabilistic_hough_line
Here is an jupyter notebook comparing the two functions on an arbitrary image (please ignore the image loading/preparation code from my project at the top, I plot the imput image "grad" under the output lines):
https://github.com/innovationgarage/price-tag-reader/blob/master/hough_cv_vs_skimage.ipynb
Thanks in advance, Egil