RAISR Sharp Images with Machine Learning

I would not compare a fancy NN algorithm to Bicubic interpolation, even as an example. Bicubic sucks. If someone really wants to get a decent high-res upscale of an image via "conventional" means (like GIMP) they would probably use something like Lanczos algorithm for upscaling -> Anthropomorphic smoothing for denoise -> unnsharp mask for clarity. And that's a simple pipeline, you can go even fanciers with masks, layers and edge-detection kernels.

Those looking for quantitative comparison of this against other techniques, it&#x27;s on page 21 of the paper ( <a href="https:&#x2F;&#x2F;arxiv.org&#x2F;abs&#x2F;1606.01299" rel="nofollow">https:&#x2F;&#x2F;arxiv.org&#x2F;abs&#x2F;1606.01299</a> )<p>Edit: and to see for yourself in more detail there&#x27;s <a href="https:&#x2F;&#x2F;drive.google.com&#x2F;file&#x2F;d&#x2F;0BzCe024Ewz8ab2RKUFVFZGJ4OWc&#x2F;view" rel="nofollow">https:&#x2F;&#x2F;drive.google.com&#x2F;file&#x2F;d&#x2F;0BzCe024Ewz8ab2RKUFVFZGJ4OWc...</a>

<a href="https:&#x2F;&#x2F;github.com&#x2F;nagadomi&#x2F;waifu2x" rel="nofollow">https:&#x2F;&#x2F;github.com&#x2F;nagadomi&#x2F;waifu2x</a><p>Here&#x27;s a previous project that attempted this as well.<p><a href="https:&#x2F;&#x2F;arxiv.org&#x2F;abs&#x2F;1501.00092" rel="nofollow">https:&#x2F;&#x2F;arxiv.org&#x2F;abs&#x2F;1501.00092</a>

This is obviously something where machine learning is a great tool, knowing about the way images usually look, it&#x27;s easy to imagine deep learning methods doing a great job at this.<p>However, an obvious application of this will be to apply these methods to security image footages, in order to improve the recognizability of the people shot. You know like in Hollywood movies where the detective asks the geek &quot;can you improve the details of that picture?&quot; But the general public (and by this I mean cops, lawyers and jury people) will not understand that the details are actually made up. The person in the security footage may be unrecognizable before treatment, but after, it would look as &quot;the most likely person to have been there.&quot;<p>What could go wrong?

Are there any machine-learning approaches to superresolution which take context into account? By that I mean filling in details that can only be guessed by context.<p>For example, when the algorithm detects that the image is of a field of grass (but individual blades of grass not discernible), then in the superresolution output it paints in these blades.

This is very clever, and an interesting read, but kind of disappointing results.<p>They don&#x27;t look <i>much</i> better than doing say... upsample to twice desired size, apply basic sharpen&#x2F;structure filter, downsample to desired size.<p>A little better? Yes. But some (like the horse) look pretty weak

I wonder how it compares with Waifu2x

Does this mean we can finally get a form of Zoom-and-enhance seen on CSI?

I&#x27;m imagining the day when old black and white films and home movies are restored to be almost crystal clear.<p>Can&#x27;t be more than 18mo away by the sounds of it.

From a qualitative glance at the images, I believe I could come close to the same results with some basic filtering, like a gussian blur followed by a quantization to a smaller number of colors, then a little bit of blur.<p>The image of the horse head statue looks like an &quot;oilify&quot; effect.<p>Mind you, a pretty good oilify effect.<p>This filter is clearly useful independently of what it achieves in relation to what it &quot;says on the box&quot;.<p>In creative image processing, I just have to like what it does, whatever that is (and I do).

Is there an executable form &#x2F; source for this we can play around with?