How do I measure density of data points within a window in plots such as the following?

Below I have a plot of (x,y) coordinate data for the trajectory of an object.

What is the best way to measure density of data points within windows of say 50 as I progress along the trajectorys path?

Thanks

[dpb]

Sounds like just binning (histogram) the X vector with desired bin size would do it for you automagically...

[James Morris]

Thank you! I'm now trying to extract data regarding the highest peaks and save them. There seem to be a few ways on MATLAB answers that people have discussed to go about it.

[dpb]

findpeaks is the quintissential MATLAB function altho it won't handle the x vector here gracefully, you'll just have to use the y vs array position.

[James Morris]

Sorry what do you mean y vs array position? Since I only used the x coordinates for my histogram I don't have any y coordinates to refer to?

[Adam Danz]

Wouldn't histcounts2() be better? The problem with binning only along the x axis is the far separation between some of those data points in the y direction. For example, at x = ~450 there are data points at y=~[350, 425] that aren't really grouped.

[dpb]

I guess it depends upon what he's really wanting -- which I'm not all that sure about. If he's following the trajectory, then binning 50 points is really just taking 50 points at a time from the beginning of the array irrespective of what the x coordinate really is. If it is actually using x, then he's got the start and the end of the trip glommed together at the first bin and that doesn't seem right on retrospect. That doesn't seem to make any sense of how to look at the data in any fashion I can think of.

[James Morris]

Sorry, let me explain it better. The trajectory plotted is the (x,y) coordinates of a hand as it enters a PIN on a keypad. The 4 clusters you can see roughly between x coordinates 500-620 correspond to keypress frames because the hand isn't in motion at those points. I want to be able to move along the array of x,y coordinates (from start to end of trajectory) and measure density over a window of say 25 x coordinates in length. Thus these clusters should then be highlighted and can be identified as the keypress frames. I aplologize, my initial idea of binning 50 points was a mistake.