I want to fill in the gaps in a matrix (extrapolation) and produce nice looking plots such as this ->

I have a plot attached (17.02.17) which was created using vectors and interpolation via binning (bindata2: http://xcorr.net/2013/12/23/fast-1d-and-2d-data-binning-in-matlab/). However, I want to create a nicer looking plot so that the image looks more smoother / higher definition (see higher def plot (17.13.33)- I didn't create).

Can someone tell me how I can do this via extrapolation (to fill in the gaps - if possible) i.e. to guess values where there are none? and also, which plotting functions are the nicest? at the moment, I'm only using imagesc and pcolor. Are there any better Matlab / external toolbox plotting functions that anyone knows of?

Here's some example code of where I am at the moment:

for d = 1:10;

      val(d).field = data_split(d).SG.field;
      val(d).y  = data_split(d).SG.y;
      val(d).x  = data_split(d).SG.x;

end

 x = [val(:).x];
 y = [val(:).y];
 field = [val(:).field];

x1rg = [7:0.020:9];
x2rg = [0:5:1000];

Output = bindata2(...
    [field],... % Value to grid
    [x],[y],... %x1 and x2 in the script, or your x and y locations in more general terms
    x1rg,x2rg);

figure;
pcolor(x1rg(1:100),x2rg(1:200),Output);
shading flat;
colorbar
caxis([8 8.3]);
xlim([7.2 8.3]);
title('title','fontsize',22);
ylabel('y');
set(gca,'FontSize',16);
set(gca,'ydir','reverse')

Answer 1

[Kelly Kearney]

A scattered interpolant is probably going to be your best bet, assuming you're starting with data that isn't on a regular grid.

x = [val(:).x];
y = [val(:).y];
field = [val(:).field];

x1rg = [7:0.020:9];
x2rg = [0:5:1000];

[x1rg, x2rg] = meshgrid(x1rg, x2rg);

F = scatteredInterpolant(x, y, field, 'natural', 'nearest');
Output = F(x1rg, x2rg);

You can play around with different interpolation and extrapolation methods; I tend to like the natural/nearest combo for data like yours that has some larger missing patches on the edges.

Answer 2

[Adam]

doc interp2

should do the job. There are examples on that page - it isn't totally intuitive to use first time. It is what I use for this kind of thing usually. I also use imagesc for plotting, I've never used pcolor not especially knew it existed so not sure what that does differently!