|Tracking Spinning μTs|| Main project page|
Previous entry Next entry
Adding to the program from the list from yesterday
The first thing i gotta do is find the radius of these objects. The problem is that most of these seem to be better described by an ellipse and not actually by a circle. There is a .vi that will fit an ellipse to the object but it needs 6 points to work. I can get these points from things like the clamp .vi. The problem with clamp is it needs a rectangle surrounding the object. That shouldn't be a problem since the match pattern gives a bounding box around the object. but it gives this box using 5 points (why i am not sure). And the rectangle needed for the clamp should be top left point and bottom right plus the angle. So i gotta write a sub.vi that transforms the bounding box to a rectangle.
Fitting an ellipse to this is a giant pain in my ass. I need 6 points to do it, and it can get royally fucked if i find bad points. this might work a lot better if i threshold it and turn it into a binary image, but that would make it harder on the user. I wanna see how easy it is. OK i got the radii through thresholding and then some magic. right now i am using an automatic threshold value this seems like a good choice but i'll see what i can do to make it more user friendly when i am done. right now i am gonna go get some lunch and be back later.
With the radii i can find tangential velocity. it just occurred to me i don't know how to find tangential velocity for an ellipse. this could be interesting.
I have successfully implemented a thresholding, then finding all points on the ellipse, and then i fit an ellipse with these points to find the radii. The automatic thresholding can be changed just not to manual yet. I will put that in and have a window pop up kind of like in my tracking software, but i am pretty sure this auto thresholding is fine.
This is made tougher since off hand i don't know the equation for this with an ellipse. it's gotta be out there though. So apparently it isn't out there. ellipses are a pain in the fucking ass and they apparently have unknown tangential velocities if they are spinning. i gotta talk to Koch about this. And see if it is necessary to describe them as ellipses.
IMAQ Rotation Detect
With the radii known now i wanna see if i can use this sub.vi to find better answers.
Ahh fuckett. it's sunday and i am feeling defeated. i don't know ellipses and the rotation detect needs the objects to be circles. it wants only one radius. I can do this of course by throwing out the ellipse shit and make them circles. but i am feeling defeated. sorry koch, i won't have this done by monday. oh yeah and the Rockies got killed in game 1. very disappointing.
OK the rox got a second game and i am back at it. This IMAQ rotation is garbage, and i don't think worth my time fixing. It basically did jack shit. it said the μT went from 3 degrees to 5 degrees while the match pattern said it went around the circle twice. So fuckett. that's done.
For right now i am going with a circle, and i'll talk to Koch about the ellipse idea later.
The tangential velocity equation is vT = ωR
To figure out the tangential i'll need ω. This is the slope of the line of angle versus time. so two things i gotta do. convert degrees into radians and make sure it adds onto 360 like i mentioned yesterday. koch linked a .vi that might help i'll take a look at that.
Koch's program looks at time, and does what i want except my circles can go in either direction. Koch looks and says if one number is larger than another do something, but i can't do that because on microtubule may go clockwise while another goes counter. I could look at the first couple and then decide if it is going counter or clock and than choose the greater than less than stuff. Goodness gracious i kept thinking that i'll do this during the program, but i can do this afterwards when all the data was collected let me think
my first attempt failed. i gotta try again. this is kind of annoying. ok i am gonna have dinner. i might be back to try some more.