I have noticed that there is a lot of aliasing when I perform downsampling (resize) or distortions on an image (warpPerspective) using any of the interpolation methods.
Is this a limitation of the opencv implementations?
Can someone suggest a method to effectively antialias the images? I have already tried the various interpolation schemes and they do not give any better results. I'm particularly interesting in correcting for the jagged edges in PNG images.
The below image has been creating using the function:
> distorted_img => cv2.warpPerspective(input_img, M,> (w_dst, h_dst),> cv2.INTER_CUBIC | cv2.WARP_INVERSE_MAP,> cv2.BORDER_CONSTANT)
**Edit: Imagemagick and Opencv Comparison**
↧