Finite difference method is used for image compression ( http://jre.cplire.ru/iso/nov12/1/text.pdf ) and ( http://elibrary.ru/item.asp?id=27147482 ) and ( http://www.freepatent.ru/images/patents/497/2500067/patent-2500067.pdf ). First formed boundary conditions (this pattern), then pattern compressed through Huffman or arithmetic coding. The method of differential compression works well on gradient images or images containing large fields of the same color or brightness. The essential question is the choice of color space YCrCb or RGB or. Also proposed ( http://jre.cplire.ru/iso/nov12/1/text.pdf ) methods of improving the quality of image reconstruction using prefiltration, with a pattern formed on the original image. In the image: a -- the original image (a photo of the Earth remote sensing); ? -- boundary conditions (pattern) in red denotes excluded elements; ? -- is the restored image. On the image: a - the original image (photo from the International Space Station); B - boundary conditions (figure) in red indicate the excluded elements; B is the restored image.
- you earned "Featured Contributor" badge, congratulations !
This is a great post and it has been selected for the curated Staff Picks group. Your profile is now distinguished by a "Featured Contributor" badge and displayed on the "Featured Contributor" board.
That looks like good restoration, Markus. Thanks for sharing it!
For comparison, see below the results of the function Inpaint with two different methods yielding similar results. You'll also be pleased with the speed.
Inpaint
Matthias, thanks for pointing out Inpaint, of course there is already a built-in function ;) I still would be curious how much one could speed up my code - probably not up to compete with the internal functions, but who knows.
