I hope this kind of how do I do things question is apropriate to send here.

I have done 3 mistakes.
1) I bought 18 000 pieces jigsaw puzzle
2) I want to learn computer vision and shape recognition
3) I'm newbie with graphics.

But I have a plan. I want to make software which cheats the puzzle. And if I get it up and running and solving puzzles I'll release it to public with source.

First I'll scan the pieces and bag them in 10 pieces/bag. Then I'll teach propably gimp how to extract the ten images from the one scanned image. That is propably the easiest part.

Next I need information of the shape of the piece. That is the trickiest part. Is vxl up to that task? And how and what tools I should use? Then I need to know the center of the piece. That is needeed when I render the picture and I have to rotate the piece.

After that it's all about number crunching and database. I'll make nice graphical application whish renders the pieces together and then user can click the definetely wrong pieces to second round. When the puzzle is solved, the application will give instruction to take next piece from bag number X.

Hopefully you know what I should do after short description of my problem. I would like links, refereces and all that kind of stuff which could help me with this project. Google brings so much theory about a lot of unrelated things, but since my laptop was able to log in with my face, I believe somewhere is something useful even for me.

Hannu Vuolasaho