User talk:TheLarry/Notebook/Larrys Notebook/2009/08/25
Steve Koch 00:23, 26 August 2009 (EDT): Yeah I was pissed to see the Manny RBI, but in the end a great result! A worry about the derivative filtering is that you don't necessarily know it's a bad point if it's a crazy value, because one bad image will produce two bad points (if frame N is bad, then frame n+1 derivative will also be bad).
TheLarry 00:38, 26 August 2009 (EDT):I don't think i follow 100%. I think 1 bad point wouldn't necessarily produce 2.
Steve Koch 00:52, 26 August 2009 (EDT):Just saying if:
- v3 = x3 - x2
- v4 = x4 - x3
Then if x3 is a wacked point, it will screw up both v3 and v4
TheLarry 00:49, 26 August 2009 (EDT):OK yeah i gotcha but both points will be screwed up. one really high the other really low right?
Steve Koch 00:52, 26 August 2009 (EDT): Yup...and say two frames in a row are bad, then the second bad frame could actually produce a reasonable-looking velocity if it's bad in the same way as the first bad frame. Maybe you can figure out how to deal with that, though...probably would require looking at the actual value of the angle at the same time -- to figure out when it gets back on track.
TheLarry 00:57, 26 August 2009 (EDT): See i wouldn't know for sure that those are bad points. i think the first bad point occurs but the second one rotates the correct amount so it would be a good point if it wasn't so high or low
Steve Koch 00:59, 26 August 2009 (EDT): Hmmm, yeah, I think we'll need to look at it together.
TheLarry 01:01, 26 August 2009 (EDT): Good idea. I think Andy and I will discuss some kinesin stuff tomorrow since i am assuming you are stuck with UNM class stuff all day and we can look at this stuff on the day after tomorrow
Steve Koch 01:02, 26 August 2009 (EDT): Actually Linh lunch party at 11:30-ish before class, and then I'm free after class until evening drinks at 7 ish.
TheLarry 01:04, 26 August 2009 (EDT): Awesome. Then we can look at it tomorrow. I hate this un-mod and removing bad point sub.vi. But I am kind of feeling this integral of derivative thing. We'll see if it actually works or not.