Excluding matrix indices which fall outside of a specified area.

Hello,

I have a dataset which I have plotted with pcolor. It is based on GPS data for an invasive plant survey (basically walking through the bush). The green indicates areas which were walked directly, the red indicates areas which were in viewing range, and the white inbetween the red and green indicate zones out of a surveyors view. The red blocks along the edge of the plot indicate the boundary of the survey area.

I want to remove the 0 values which fall outside of the survey area (these are the 0 values which continue to the grid's X and Y limits). What I cant wrap my head around is how to remove (or NaN) only the zero values which are not surrounded by data values of 0.5 or 1 (does not need to be a neighbouring grid block).

I want to remove these values so that I can work out the total area which the survey covered compared to the total area of the survey plot (I havent started this yet).

I have attached the pcolor figure and the code block which created the plot.

Thank you in advance,

Nic

Answer 1

[Mathieu NOE]

hello

this would be my suggestion ...

n = 50;
Z = peaks(n); %returns an n-by-n matrix.
Z = abs(Z);
Z = Z./max(max(Z));
Z(Z<0.33) = 0;
Z(Z>=0.33 & Z<0.66) = 0.5;
Z(Z>=0.66) = 1;

figure(1);pcolor(1:n,1:n,Z);colorbar('vert');

[FX,FY] = gradient(Z);
eq_grad = abs(FX)+abs(FY); % my special equivalent gradient
eq_grad = eq_grad./max(max(eq_grad)); % normalized to 0,1 range
% figure(2);pcolor(1:n,1:n,eq_grad);colorbar('vert');

% create NaN where eq_grad is below threshold and Z < 0.5
ZZ = Z;
thr = 0.001; % gradient threshold slightly above zero
ZZ(ZZ<0.5 & eq_grad <= thr) = NaN ; 
figure(3);pcolor(1:n,1:n,ZZ);colorbar('vert');

[Nicholas Salonen]

Ah that doesn't quite do what I need. So if you were to imagine a boundary which circles around the peaks which you made, I need to retain those zeroes and make NaN all the values which would lie outside of this circle.

i.e. see attached.

[Mathieu NOE]

hello again

so this is the new code and its output

hope it helps

n = 50;
Z = peaks(n); %returns an n-by-n matrix.
Z = abs(Z);
Z = Z./max(max(Z));
Z(Z<0.33) = 0;
Z(Z>=0.33 & Z<0.66) = 0.5;
Z(Z>=0.66) = 1;

figure(1);pcolor(1:n,1:n,Z);colorbar('vert');

[FX,FY] = gradient(Z);
eq_grad = abs(FX)+abs(FY); % my special equivalent gradient
eq_grad = eq_grad./max(max(eq_grad)); % normalized to 0,1 range
figure(2);pcolor(1:n,1:n,eq_grad);colorbar('vert');

[y,x] = find(eq_grad == 0.5);
shrink = 0;
k = boundary(x,y,shrink) ;%      % Repeat this example by separately passing
%      % a shrink factor of 0 and then 1 to the  boundary function.
xb = x(k);
yb = y(k);

hold on;
plot(xb,yb);
[xq,yq] = meshgrid(1:n);

[in,on] = inpolygon(xq,yq,xb,yb);
plot(xq(in),yq(in),'r+') % points inside
% plot(xq(~in),yq(~in),'bo') % points outside
hold off;

% create NaN for outside points
ZZ = Z;
ZZ(~in) = NaN ; 
figure(4);pcolor(1:n,1:n,ZZ);colorbar('vert');

[Nicholas Salonen]

Thank you! Will adapt to my work!

Cheers

[Mathieu NOE]

you're welcome