User:TheLarry/Notebook/Larrys Notebook/2009/08/15

{| width="800"
 * style="background-color: #EEE"|[[Image:owwnotebook_icon.png|128px]] Tracking Spinning μT
 * style="background-color: #F2F2F2" align="center"|  |Main project page
 * style="background-color: #F2F2F2" align="center"|  |Main project page


 * colspan="2"|
 * colspan="2"|

Back to the old days
I wrote a color tracking software trying to keep track of the spinning microtubules, but that doesn't seem to work that well. There are some options to play with but none seem to get what i want. I can try to get creative, but for now i am gonna drop down and try to do this in black and white and see what happens. If that doesn't seem to get me anywhere, then i'll look online for help.

I was playing around with the color before but now i AM going to write a black and white search program. I did think of changing the range of angles that the .vi searches over -- this could help a bit.

There are some advanaced .vis for black and white. They have more options than the unadvanced versions. Right now i am programming with unadvanced but this can change when i get the structure down. or if the unadvanced doesn't work

http://www.youtube.com/watch?v=ANWD8-u7-No

   
 * Andy Maloney 18:31, 15 August 2009 (EDT): Make sure you put the html tags around it like in the wiki code. Why didn't you put this on the KochLab YouTube Channel? Very cool. If you do a text search for the word "embed" on the YouTube link, you will find the code to embed the file.
 * TheLarry 19:06, 15 August 2009 (EDT): Hey Andy, thanks for the help. I just didn't want to spend too much time wading through it. I'll put this on koch lab when its done. It just ain't there yet.

there's the link. I can't seem to embed it. I am looking for the id= part but it ain't there so i'll just put the link in. the black and white works a lot better than color. I still have some options left. like changing the angle range -- this might just make it faster which isn't a problem. I can also add in advanced stuff but i am not sure what that'll do exactly.

All of the advanced stuff seems to give more accuracy to the answers. So this is probably something i'll want to add. There are two .vis. IMAQ Advanced Setup Match Pattern 2 and IMAQ Advanced Setup Learn Pattern 2.

Also i might want to add the ability to track more than one at once

Steve Koch 00:49, 16 August 2009 (EDT): That's great! Looks like it's working really well...and I swear from the video that that circle moves (translationally). Is that true?

Improvements and Additions

 * 1) I gotta keep track of how many times the object that is spinning crosses 360 (or 0). Once it goes from say 360 to 380 the program says 360 to 20 i gotta make that 380.
 * 2) * Steve Koch 00:47, 16 August 2009 (EDT): I have a VI for the feedback that keeps track of when a 24-bit clock flips over--probably this would be a similar thing. http://kochlab.org/files/Software/Feedback96--First%20version%20that%20compiles/Modules/CalculateClockDiff_24bitRollover_2009.vi
 * 3) *The problem is that bad points can fuck up whatever way i choose to do this.
 * 4) I gotta write something to calculate the tangential velocity
 * 5) *For this i'll need the radius
 * 6) save data
 * 7) * Steve Koch 00:47, 16 August 2009 (EDT): I'd say tab-delimited text files would work well for her.
 * 8) add in advanced sub.vis
 * 9) *I'll have to play with the many options this creates to see what works best
 * 10) add possibility to track multiple circles at once
 * 11) *this could be toughest thing on this list
 * 12) * Steve Koch 00:47, 16 August 2009 (EDT): It sounds slick, but maybe's not necessary if it's that hard. In the "circles analysis" VI that I wrote for her, I dealt with multiple ROIs.  Perhaps that would help.
 * 13) limit angles that the .vi searches through.
 * 14) *this is a speed up technique and i don't think is necessary
 * 15) IMAQ Rotation Detect

Radius Figure Outting
I think the clamp sub.vi might be able to help me with this.

There is a sub.vi called IMAQ Rotation Detect. This might give better answers than the pattern match, but it needs a radius to work so i can add this in after i figure out the radius.

The clamp is annoying since i need to give it a rectangle input which is different from bounding box. so i gotta make a sub.vi that makes this

I am not sure why but it became important to me that i don't just find one radius but two since most of these look like ellipses.

Turning this bounding box into a rectangle is more annoying than i am ready for and i am a bit hungry. so i'll have to work on this tomorrow or later tonight. we'll see