I am trying to reproduce a result from OpenCV using SkImage. In OpenCV we have:
img1 = cv2.resize(img, (w, h), interpolation=cv2.INTER_AREA)
In SkImage, my best shot so far is:
img2 = skimage.transform.resize(img, (w, h))
However, I am still not able to get the same result as OpenCV. I read the documentation on cv2.INTER_AREA and this article, but I am still unsure of how to reproduce this result in SkImage.
What is missing from the skimage implementation? A PIL.Image implementation could work too. Suggestions on how to diagnose the problem are appreciated.
Examples of the difference between img1 (OpenCV) and img2 (SkImage) are as follows:
You are using the width and height in skimage
in wrong order. It should be reversed like below:
img2 = skimage.transform.resize(img, (h, w))
Then compare the result. Usually skimage
produce better result compared to OpenCV.