Date: Mar 20, 2014 3:18 PM
Author: Kelly Kearney
Subject: Re: Colormapping Matrix with 99-90% zeros and several peaks

"Julian" wrote in message <lgfdru$p68$1@newscl01ah.mathworks.com>...
> I have a Matrix with values ranging from 0 to 2*pi, size of this matrix is 2000² (just as a reference) and there are very few values in there (a maximum of 2*32²).
>
> Now i need a method or an idea to visualize this. What I have so far:
>
> figure;
> image(Z1) %Z1=matrix<=size=2000*2000
> colormap(jet)
>
> Of course Finding those pixels is hard and they aren't a very good representation.
>
> So my thougth was to make a convolution of the matrix with a 2D-gaussian curve (curve beeing normed at maximum to 1).
> But that seems a very crude methode and I have to use this a lot.
>
> Furthermore I'd like it to exclude all zeros ( colormap from white (0) to blue(lowest) to red (highest)).
>
> I really hope for some ideas or usefull matlab methods.
>
> Thanks in advance


I think a scatter plot would be your best bet. This function

http://kellyakearney.net/matlabdoc/GeneralUtilities/spycolor.html

is a little wrapper I wrote that creates a scatter plot for just this type of sparse data (and it will exclude 0s by default).