0
$\begingroup$

I have 3 versions of an image at lower resolutions - 480p, 720p, and 1080p. Is it possible to use those 3 images to predict the next highest resolution image (the 4k image)?

My first guess was that I could calculate the difference (or momentum or acceleration?) of each pixel between images and use that to predict the new 4th image. I assume this would require me to give all pictures the same number of pixels. I also assume that the differences in each pixel between images would need to be smooth. Would this work?

My other guess was that maybe there is some kind of statistics or machine learning technique that I could use, but I don't know anything about that stuff.

Thanks!

Ps - this man is not my real data. My real images are an intensity map (temperatures over an area) at different resolutions. I suspect that the introduction of colors might complicate things, so should we pretend that the images below are black-and-white images?

Here are the pictures: Luther at different resolutions

$\endgroup$
  • $\begingroup$ so, are the lower-resolution images independent observations, or smaller versions of the same image? $\endgroup$ – Marcus Müller Aug 6 '17 at 22:22
1
$\begingroup$

There is no extra information in the 720p and 480p images that is not already in the 1080p image. You can do the interpolation on the 1080p image. A recent work that may be of interest to you is Ledig et al. Photo-Realistic Single Image Super-Resolution Using a Generative Adversarial Network, arXiv:1609.04802v5, 15 Sep 2016 (v1), last revised 25 May 2017. But if your work is for scientific purposes rather than for looks, then it is probably best to use traditional interpolation methods that do not try to increase the resolution by making up new data.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.