First week

1. I compiled GDAL and implemented simple prototype of gradient descent method for searching
   and optimization of transformation parameters for 2 photos. Of course, it's far away from perfect,
   but it looks promising.
2. I think that I'll start to learn SURF algorithm, how it works, maybe I'll begin to code prototype to verify my awareness. It's a very sufficient part of stitching processing.
3. Usually it's very hard to start something. You have to dive into details and get used to work
   with many new things. I have lots of homeworks in my university study and exams are coming.
   So for now it's extremely exausting task to mix SoC coding and study.

Cool example of stitching results at this moment.
Here are 3 iterations - initial, intermediate and final. White lines are distances between corresponding points on both images. With every step, algorithm moves points (images) closer to each other, so distances (errors) become extremely small.

Few days ago I worked on gradinet descent method for optimized stitching. I wrote the dirty implementation on python + numpy. I've done some primitive tests and verified in C# little program, which I made for viewing points before and after optimization. I've discovered that algorithm converges, and even it requiers sometimes lots of time (but still admissible), it seems to me that I'm on right way.