User:TheLarry/Notebook/Larrys Notebook/2010/01/27

From OpenWetWare
Jump to navigationJump to search
Tracking Software Main project page
Previous entry      Next entry


I think it is possible to automate this tracking software. Not the smoothing but i think it might be possible to run the tracking software overnight and track every single microtubule that appears in every folder of .pngs that you tell the software. This is definitely going to be tough. Just to get the idea down and then to actually implement. But i think it is possible.

My idea right now is to take all the center of masses in the video (do this through the particle analysis .vi) and then check to see if the center of mass is in which bounding box from the previous image. If it's in more than one then check the particles' information (area, perimeter, etc). The object with the smallest difference in these characteristics from the previous frame should be the object in this frame. I don't know if this'll work. I have to think about it.

The main object is to write a that outputs starting and ending frame of every microtubule without looking at the same microtubule twice, and following any new microtubules that pop up. I don't think the above will work, but it might be on the right path.

OK I have to rethink this. But I can do that at home. It is interesting that the above idea means tracking all the microtubules which i do later in the program. There has to be a simple way to do this...but it might involve keeping track of a lot of data which i am not good at. If it's possible i'll find a way to do it.

OK another way that might work, but take a long time would be...ok imagine the tracking software runs on the first particle in frame zero. And then it tracks particle 1...till it tracks all the particles. I don't know how to cut it off yet, but i'll think about that later. After it tracks all particles that are apparent in frame zero it goes to frame 1. Now it checks each particle there. What i mean by check is that it will pattern match that frame and if it finds the ends match a particle from the tracking from frame zero it won't track that particle. So most of the time it won't track anything. This might take a while since there are so many .pngs but if this can run overnight then speed won't be much of an option. I'll look into this tomorrow.