Increase recursive loop performance

Hello guys, My code works with recursion, receiving an image as input, but it's taking a really long time to process it. I wonder if is there anything I can do to speed it up.

Here is the code:

                 ROTULO=0;
                 for i=xi:xf
                      for j=yi:yf
                          if(out(i,j,1)==-1)
                              ROTULO=ROTULO+1;
                              rotular(i,j);
                          end
                      end     
                 end

And the function rotular:

function rotular(w,z)
out(w,z,1)=ROTULO;
if(w < xf && out(w+1,z,1)==-1)
    rotular(w+1,z);
end
if(z < yf && out(w,z+1,1)==-1)
    rotular(w,z+1);
end
if(w > xi && out(w-1,z,1)==-1)
    rotular(w-1,z);
end
if(z > yi && out(w,z-1,1)==-1)
    rotular(w,z-1);
end
end

The variable out is a matrix corresponding to the input image. I'm running MATLAB R2015b on Windows 8.1, 64-bit, 6.0GB RAM, Intel i5-3470S CPU @ 2.90GHz I'm not having any trouble concerning RAM. I am also not getting any errors. Just want to improve the performance.

Thanks.

Answer 1

[Walter Roberson]

In MATLAB, passing parameters explicitly is a little more efficient than shared variables, so it would be a little more efficient to pass in ROTULO and to return "out" from rotular

Shared variables are not so bad -- much more efficient than global -- but explicit variables are faster.

Also, R2015b apparently improved the efficiency of recursive routines.