I got back from the IMA workshop on ‘New Mathematics and Algorithms for 3d Imaging’ last night. It was not what I expected, in that the most represented areas of imaging were the ones dealing with tomographic reconstruction and the Radon transform. I had been expecting it to be geared more towards wavelets and using other image representations to form efficient algorithms for image processing– so I learned a lot more about the Radon transform than I had anticipated. This was a good thing. It also defied my expectations in other ways: PDE methods, like Mumford-Shah and active contour methods were also well-represented. In fact, wavelets, curvelets, etc. were the least represented areas. Also, physics was remarkably prominent, in the form of the wave equation, diffraction, scattering, and such. So it seems that I should learn about PDEs, physics, optimization, and such things. I have a large list of key ideas to look up when I get the time.
Unfortunately, it seems I will be starved of time this semester. In addition to taking several hard classes, I will be working two jobs, and I still don’t have a car. The two classes that have me worried most are my senior design class– the final, and very time consuming requirement for me to get my EE degree– and a Fourier course I’m taking with Papadakis based on Grafakos’ ‘Classical and Modern Fourier Analysis’. Just looking at the second page of the first chapter in that book is scaring me. This weekend I’m going to try to read through as much of it as I can understand, so I can manage not to embarass myself in front of the graduate students; I’m sure I’ll have to refer to references often, so I need to keep way ahead of the class. Luckily there will be no homework in there, although I will try to do a couple of the exercises every now and again.
My last problem: applying to grad school. I still haven’t finished that. I’m only applying to UH, CalTech, and the University of Maryland. CalTech’s is due on Sunday, so I’m trying to finish that up today. Next I’m going to complete Maryland’s due Feb 1st, and then UH’s. Once I get those out of the way, that will be a load off me, and I’ll be able to finally fully concentrate on school.
Possibly relevant posts:
I’m still in the office at school… but at least I finished my poster for the IMA workshop. I wrote it in LaTeX, using the sciposter class– I had started using a0poster, but then realized how much more convenient sciposter is. It was just like writing a paper; in fact, I could just change the documentclass (ok, and a few other minor changes– mostly commenting out lines) and get a report. In fact, I think I’ll do so, just as a proof of concept. Neato, ain’t it? Well, except for the fact that two of the images in the subfigure environment have vanished. But let’s not be so small-minded.
Here’s the whole shebang– Kile project, graphics, bibtex, LaTeX code– if you’re interested in making your own LaTeX posters.
Possibly relevant posts:
- A Possible Future (6/3/2005)
- jsMath (4/20/2005)
- Barycenter (9/23/2004)
My new year’s resolution: to effectively manage my time. Not to learn how to do it– what’s to learn?–, but to implement it. No more procrastination, or reading of novels instead of textbooks: delayed gratification is the name of the game.
Why is this my resolution? It’s been on my mind since my freshman year in high school. I’ve screwed up several aspects of my life by refusing to do things when they should be done, and it’s become an ingrained habit. So for the next year, the painful process of ridding myself of this habit is going to be my top priority. I can’t get anywhere in the academic world if I don’t beat this out of myself.
Possibly relevant posts:
Have you ever wondered what algorithm is used in visualization interfaces, which allows you to rotate an object in three dimensions in a natural manner using your mouse? I have, and I’ve been somewhat shamed by my stupefaction: it seems like such a widespread interface element, available in even the crappiest of 3d applications, that I thought I was overlooking an obvious solution.
Well, I have a vindication of sorts. In compiling some OpenGL demos, I stumbled across the file trackball.c, which implements such an algorithm. From glancing through the code, it seems that it isn’t such a simple problem after all, and there’s no one obvious answer. This particular implementation, which from the familiar feel of the results is a standard one, is based on a projection onto a virtual trackball followed by a hyperbolic deformation— or is that a hyperbolic rotation, which results in a deformation? Whatever. Reading code comments can get you nowhere fast. Luckily the people who wrote the code began with a SIGGRAPH reference:
… lots of ideas from Thant Tesman and the August ‘88 issue of Siggraph’s “Computer Graphics,” pp. 121-129.
So, that’s something for me to look up. In fact, I’m sure the method has been improved upon, so I should probably look for works which reference that also.
Possibly relevant posts:
Happy New Years Eve!
I spent all of today getting the family computer ready for a fun filled year of programming. That meant downloading lots of software, and installing about 10% of it. It also entailed rediscovering why I stopped trying to program Windows code a long time ago– the same reason I despised C for such a long time: the black magic you have to know to work with nontrivial software packages, especially libraries. I used to think installing software under Unix was a nightmare, but compared to installing libraries for use with Mingw32, that’s a cake walk.
As of right now, I have reached a logical stopping point: I have gtk, freetype, glut, and gtkglext installed and working with my Mingw32 setup. Once I can install agg– it has resisted my first attempts– and learn how to use it in conjunction with gtk, I’m all set for several projects I have planned. I’m also looking at wxWindows as an alternative to gtk, mostly because that came available as a DevPack.
By the way, the best way to get Mingw32 is definitely as part of the Dev-C++ distribution. So far, the IDE has earned nothing but praise; in the short time I’ve been using it, I’ve already found the ability to import Visual C++ .dsp project files invaluable, and I’m sure I’ll have occasion to use its autocompletion facilities. Another neat feature it has allowed me to install the Win32 api, OpenGL, and Gtk help documents in the help interface. But, the nicest feature of Dev-C++ would have to be the builtin support for DevPacks— archives of code that can be downloaded and installed from within Dev-C++ –, which simplified my life considerably.
I look forward to screwing about with Mingw32, OpenGl, and various windowing kits throughout the coming year. Who knows? If I get really into all this, I might even take the time to learn the dreadful Win32 Api.
Possibly relevant posts: