Intelligent Image Resizing

Few weeks back, I had writ­ten about Pho­to­swap­ping images instead of Pho­to­shop­ping them. Here’s anoth­er ground­break­ing algo­rithm for “con­tent aware image resiz­ing”, thanks to Michael from Techcrunch:

The research paper, as well as a down­load­able video, by Dr. Ariel Shamir and Dr. Shai Avi­dan is avail­able here.

Note the dif­fer­ences between the ear­li­er algo­rithm and this one. The ear­li­er one used a large data­base of images to add or replace image sec­tions. This one doesn’t work with any exter­nal images, and is pure­ly an algo­rith­mic advance towards high­er con­tent aware­ness of the exist­ing image itself. Now you can eas­i­ly use pho­tos from your Ibiza vaca­tion to print fam­i­ly-friend­ly pho­to albums!

It is fas­ci­nat­ing to see how we’re mak­ing advances in algo­rithms even today — why wasn’t this invent­ed all these years that Pho­to­shop has been around? (Sim­ple answer: because no human mind had con­ceived and invent­ed such an algo­rithm before).

I do wish the inven­tors get rich­ly reward­ed for this mag­nif­i­cent piece of work!

Tech­no­rati Tags: , , , ,

Share this post :

This entry was posted in Arts, photography, Science, technology. Bookmark the permalink.
  • This is bril­liant!
    Er, what is it, actu­al­ly?

  • Wow! This is indeed bril­liant. Thanks for high­light­ing it!

  • Arun: thanks!

    Ram­bodoc: I pre­sume you’re jok­ing…:-)

  • This is awe­some. It would hours to do some of the stuff he is doing on Pho­to­shop

  • No, seri­ous­ly, this looks great and all, but what would I use it for?

  • Karthik Krish

    Its not as fast as the videos show it to be. There is a lot of pre­com­pu­ta­tion done beforehand(the ener­gy map for exam­ple). The idea is cool but the per­for­mance on images with a lot of edges(high ener­gy) needs to be seen.

  • doc — the guy men­tions it at the begin­ning. Basi­cal­ly to allow for “more effec­tive rep­re­sen­ta­tion of the entire scene” in a small­er ver­sion which could be used for small­er dis­plays like say on an iPhone ;), and also allow for gen­er­al web con­tent (to acco­mo­date slow­er con­nec­tions bet­ter)

  • Ram­bodoc: Regard­ing what is it: have you ever tried resiz­ing an image in any image edit­ing tool? If you resize it in a diag­o­nal direc­tion, you can keep the aspect ratio intact and there’s no prob­lem. But if you stretch or com­press an image in only one direc­tion — hor­i­zon­tal­ly or ver­ti­cal­ly — it gets skewed and dis­tort­ed.

    This algo­rithm is the first one to let you stretch or com­press an image in one direc­tion with­out dis­tort­ing it. It ‘intel­li­gent­ly’ fills in the emp­ty space cre­at­ed while stretch­ing, and merges the bor­ders ‘intel­li­gent­ly’ while com­press­ing. That’s how you can turn the pho­to­graph of a cou­ple into that of a sin­gle per­son or expand the sky in a pho­to­graph for a bet­ter com­po­si­tion.

    Regard­ing what you would use it for, I don’t know. I’m sure image artists will use it for a lot of the behind-the-scenes image edit­ing that we are not even aware of when we browse the web, read mag­a­zines with cool pho­tographs, and buy posters or greet­ing cards with beau­ti­ful-look­ing images. Your ques­tion is a tricky one: it was asked by mankind when some­one first invent­ed the wheel. It was what made the Prod­uct Man­ag­er for the IBM PC say that “there is a world mar­ket for maybe five PCs”. 🙂

  • Ah, now the post makes some sense. After all, the last refuge of the edu­cat­ed moron used to be the Med­ical school, so you know why I ask these ques­tions! 🙂

  • Ah! How blessed I am to be in such enlight­ened moron­ic com­pa­ny! 🙂

  • That looks real­ly cool. I can see how this can be help­ful, but wow, did that go way over my head. way, way over my head. But, thanks for shar­ing.