Post-Processing

Hello everybody

Kellenjb

Hello

Rebecca Chernoff

8:49 PM

Allo! (:

9:15 PM

Hello! = )

genesis

hi :p

Hi all

I wonder if any follow ups on my paw questions would fit the site

paw?

genesis

paw?

Q: Peak detection in a 2D array

274

I'm helping a veterinary clinic measuring pressure under a dogs paw. I use Python for my data analysis and now I'm stuck trying to divide the paws into (anatomical) subregions. I made a 2D array of each paw, that consists of the maximal values for each sensor that has been loaded by the paw over...

88

Q: How can I improve my paw detection?

After my previous question on finding toes within each paw, I started loading up other measurements to see how it would hold up. Unfortunately, I quickly ran into a problem with one of the preceding steps: recognizing the paws. You see, my proof of concept basically took the maximal pressure of ...

python image-processing

9:25 PM

Ivo asked about his paw quastions

@IvoFlipse they would.

Lol. I thought paw stood some some algorithm name or something = )

But Ivo you can't have a diamond on this site too :p

Yes, I believe they would

@jcolebrand Why not? ;P

if all goes well I should get a project to keep working on it for 3 years :P

I was torn between asking it here, on SO or on Stats.SE

9:27 PM

@IvoFlipse :p cos I want to have more'n one too

@jcolebrand I count one, two, that leaves 1 for someone else ;P

What's a diamond anyway?

@Phonon More responsibility mostly, so its not even fun

@jcolebrand only issue is that I won't have much time to actually program/test anything before my vacation in two weeks :(

would hate it if I can't follow up on my question

Does anyone here actually work in the signal processing field? Or are we all mostly coders who have to do DSP once in a while?

and it sucks I couldn't compile the Kinect driver for Python, else I could have asked a question about that

@Phonon define signal processing ;)

9:35 PM

@IvoFlipse lol

@IvoFlipse that does suck. :(

@Phonon I'm a coder with a degree in EE.

ok, places to be. Gone

@IvoFlipse Don't even go there. It's the end of the day. = )

I do get your point though

@Phonon well I just mean, I need to analyze pressure measurements, which in itself is also just a signal. But am I a programmer or an electrical engineer? nope

Cool.

and what is it you all do?

I'm an audio electrical/DSP engineer, recently out of grad school.

9:49 PM

@Phonon You could always use your exam questions from your master year as 'seed' questions ;-)

lol

I've got plenty of my own, trust me = )

@yoda in what kind of field if I may ask?

@IvoFlipse Oh, I'm not "working" working :) I'm a grad student :)

@yoda well what are you studying then?

@yoda hmm that's a little over my head, but it does sound cool

in a weird Discovery Channel kind of way

@IvoFlipse haha, well... mostly over my head as well :)

10:04 PM

what year are you in?

i've finished yr 3 of my Ph.D... can't say how many more though :D

hello @Jonas

Shouldn't that normally be around 4?

Hello @yoda

Q: How can I rotate a 3D array?

I'm planning on asking a question like this one

6

Currently, if I want to compare pressure under each of the paws of a dog, I only compare the pressure underneath each of the toes. But I want to try and compare the pressures underneath the entire paw. But to do so I have to rotate them, so the toes overlap (better). Because most of the times t...

python image-processing

@IvoFlipse around 5ish in the US

10:06 PM

where I need to find a way to automagically rotate the paws based on certain attributes within each paw

Oh, the dog paw question. I put that as a great sample question on area51

@Jonas why thank you :)

@IvoFlipse: so what you want is image registration?

@Jonas I don't really know the right terminology, since I'm not really familiar with it

this site has great potential to become a good hub for serious discussion + theoretical concepts + some indulging of newbies like math.SE or cs.SE... I hope it doesn't devolve into a can I haz the codez for my microcontroller

5

10:08 PM

@yoda that's what moderating is for

though it would be useful if users rather than directly downvote, try and guide questions into the format we want here

@IvoFlipse: Would this help: ieeexplore.ieee.org/xpl/freeabs_all.jsp?arnumber=1643701?

oh hey tomwij answered your paws Q

> image cross-correlation, that part sounds right yeah

@yoda that was a manual rotation, I first had to derive the angle from the location of the toes... which well isn't uber reliable to begin with

@yoda: what is random matrix theory?

@Jonas in a nut shell, it means that the eigenvalue distributions of certain random matrices are deterministic

10:16 PM

Clever

for example, try x=randn(1000);hist(eig(x+x'),20)

the distribution will look like a semicircle

no matter how high you crank up the matrix size or how many times you try it, it'll look the same...

Wow

It's a miracle! :)

@yoda I like that, because it means our pressure plate doesn't have to be 100% accurate as long as the relative differences are equal

I like it if results come out to pretty

Would this also allow me to test for randomness of an array?

@Jonas have any examples of that technique being used on images?

10:25 PM

@IvoFlipse: The Theriot paper? Basically all their keratocyte literature after that. I never needed to correct for rotation that much, but this is what I'd implement if I ever had to.

@Jonas hmmm... I'm not sure. It usually is helpful in determining if outputs from different array elements are uncorrelated

Usually your signal is correlated and noise is not.. so you use this distribution to fit the eigenvalues due to noise, which you then throw out. Thus you reduce the dimensionality of the system to keep only the signals you want

@yoda I wonder if that would help me sort my paws too, because the location of the toes should be strongly correlated between contacts of the same paw

@yoda: I'm testing a new CCD camera soon, and one question is whether the camera itself introduces some noise correlation

@IvoFlipse that's true, I should read through your paw questions in detail today.. it might be really interesting if we could (and there might be a publication in there if it's possible)

@yoda I'm all ears, though I actually have to focus on the Town Hall Chat :P

10:30 PM

@IvoFlipse oh btw, quick Q.. how many elements does your sensor have?

@yoda the plate is a minimal of 64x64 sensor with length of 7 (l) mm x 5 (w) mm. In this case it was a long version with 4 after one another, so 256 x 64

@yoda on average though most paws are 10x10 pixels/sensor which isn't really high res

@IvoFlipse 10x10 might be a little too hard, but 64x64 or 256x64 is pretty good actually

@yoda: I may put that as a question onto the site

time for dinner. See you guys.

@yoda also the data is measured at multiples of 125 Hz, up to 500. In this case it was 10x10 or 15x15 pixels for about 50 frames (at 125 Hz)

@yoda cheating by interpolating the data seems to help in other studies on human feet, though I'm not sure why

are fMRI techniques on-topic for this site?

or more specific: statistical parametric mapping and the like

That's another thing that might be interesting to look at for this site

@IvoFlipse I don't see why fMRI or statistical parameter mapping etc would be off-topic

There might be concerns that the latter might be better suited for stats, but that's to be decided on a case by case basis