(Date: 12.8.2013 – 16.8.2013)
I’m still thinking hard about my multipeak finding algorithm. Trying different filter masks (see last posting) or other ideas. The Laplacian filter mask from the last post does indeed sharpen peaks but results in a histogram of second derivative multiplicities. So it just shifts the task of finding peaks in the Hough space towards finding peaks in space of Laplacian-transformed Hough space points. Do’h. During this process, peaks are emphasized, that’s true – but not the way I was hoping they would be.
I have an idea about using some fancy circle criterion with weighting factors amplifying diagonal elements. A first, messy implementation results in a histogram which has successfully been cleaned from all the original peaks and now has new ones at strange positions. Though I don’t understand yet what’s happening there, it might be something to exploit. Or to fix.
You see: I’m experimenting with algorithms. Probably I will read up a bit more in my image processing literature.
I presented my problem to the group on our weekly meeting this week. You find the slides at the end of this posting.
In the forum for our computing environment PandaRoot, I pose the question why the program re-sets the
$PATH shell variable every time it’s called.
Background: There’s a
config.sh file being built during compilation which sets some necessary environmental variables. Among those, at the very end of the script, the
$PATH is set:
Yes, you see that correctly. The whole
$PATH is overwritten to the value it had during compilation of PandaRoot. And it’s quite long (you can scroll…) and arbitrarily messy.
I didn’t like that (the overwriting, not the messiness) and asked in the PandaRoot forum, why this overwriting is needed — and why an usual $PATH extension ( export PATH="/path/to/PandaRoot/bin:$PATH") would not suffice. Well, after some discussion there’s now a cmake compiler flag which sets the export PATH="/path/to/PandaRoot/bin:$PATH" according to my suggestion: cmake .. -DUSE_PATH_INFO=TRUE.
On Thursday there is an institute’s information meeting about a new green house being constructed right next to our institute’s building. They explain to us the motivation for doing so, how it’s going to look like and what the course of construction is going to be like. And they open themselves for discussion. Which was quite, well, let’s say, interesting.
Anyway: The small forest thingy in front of my window is being replaced by green houses by beginning of spring next year. For science. Could be worse.