Wow, I’m just in time to wish you a happy new year !
Once again this month, I’m right on the deadline: I did finally hand out my thesis to the jury right on time a few days ago. Now, I’m waiting for the exam at the end of February.
A scoop: the title reads Patch-based methods for variational image processing problems.
Oh, and it’s mostly about non-local image processing, plus a chapter on Local Binary Descriptors reconstruction.
2013, a new beginning
Finishing the thesis was not easy. Part of the difficulty, I’ve officially left the lab and started a new career as a DSP engineer with the beginning of the year. I’ve discovered the wonderful world of electronics, analog and digital hardware, realtime1 processing, and embedded processors where a byte can have more than 8 bits.
2013, year of the snake
Part of the new year resolutions, learning a new programming language! Since 2013 is the year of the snake, it seems a perfect time to learn Python ;-)
I started using matplotlib to generate the figures of my thesis, and I was quite impressed. I had dozens of csv files with experimental results, and thanks to the excellent online documentation I was able to read, average, plot my files in 30 minutes. A great piece of software, and kudos for the amazing documentation!
And by the way, unlike matlab, matplotlib is able natively to write your figures to pdf files that are correctly formatted. This makes you save a lot of time, quite handy when it’s about time to release your thesis manuscript…
For my thesis, I did a small experiment to compare a semi-local vs. a non-local version of NL-means. The results were not unexpected (in my point of view), but they are probably counter-intuitive and worth a post soon.
- Actually, much faster than that. ^