Aligning images

Image alignment has quiet a lot of applications, for instance for creating panoramas. Or when creating HDR images from a bracketing when images weren’t perfectly taken from the same position. And I want something simple, one click and done. Hugin is a good tool for creating panorama, but you need a degree of computer science with a major in panorama creation to achieve a good enough result.

It’s a feature I wanted to have for 1.6, but at the same time I started to work on 1.6 features, my PHD director ask me to work on a similar stuff, and working on the same thing during the day and during the evening is not something I am capable of doing. So I left some unfinished code sleeping for almost a year and decided to resurrect it last week. The two images below were automatically aligned:

The problem is that the pictures are slightly distorted, that’s why they don’t perfectly blend into each other. This need to be corrected, which is the next step I will be working on.

This entry was posted in Krita, Open Source and tagged , , , . Bookmark the permalink.

5 Responses to Aligning images

  1. Pavel says:

    Hi, I worked on medical image registration for the past two years, perhaps you will find some of my code useful, you can get it here: http://www.sci.utah.edu/~koshevoy/research/ Paul.

  2. Cyrille Berger says:

    sounds interesting, I will have a look at it.

  3. Pooh22 says:

    Hi Cyrille, do you know that the Hugin developers are thinking of moving the UI to qt4? Also the latest beta has the ability (though I haven’t tried it myself yet) to correct Chromatic aberations by aligning and distorting each individual colour channel.There’s actually a google SOC proposal for the UI rewrite/redesign..

  4. arno says:

    great, I won’t even have to bother aligning my photos from Rome !From an end user perspective, it’s great news, thanks Cyrille

  5. sadiq says:

    Hi Cyrille,You might want to take a look at:http://www.ces.clemson.edu/~stb/klt/Both the algorithm and public domain library are quite good for feature tracking across two images. With enough features tracked, you would have enough information to possibly correct the distortion.Good work!

Leave a Reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <s> <strike> <strong>