How to extract the the right X,Y,Z value for obtaining the best contour plot similar than the complete contour

If I have a matrix A that contains 1041 rows with 3 columns (X, Y, Z)

How to extract exaclty 424 row-values from A that fits the best contour plot similar than the contour with all the rows values?

find attached A matrix [1041,4]. I would like to obtain the B matrix [424,4] still represent the best way the contour plot of:

x = A(:,1);
y = A(:,2);
z = A(:,3);

Answer 1

[KSSV]

Option 1:

load("A.mat") ; 
x = A(:,1);
y = A(:,2);
z = A(:,3);

m = 21 ; n = 21 ; % can be changed 
xi = linspace(min(x),max(x),m) ; 
yi = linspace(min(y),max(y),n) ; 
[X,Y] = meshgrid(xi,yi) ; 
Z = griddata(x,y,z,X,Y) ; 
idx = ~isnan(Z) ;

iwant = [X(idx) Y(idx) Z(idx)] ; 

Option 2:

load("A.mat") ; 
x = A(:,1);
y = A(:,2);
z = A(:,3);

[c,ia,ib] = unique(y) ; 
N = length(c) ;
iwant = cell(N,1) ; 
for i = 1:N
  xi = x(ib==i) ; 
  yi = y(ib==i) ;
  zi = z(ib==i) ; 
  %
  n = round(length(xi)/2) ; % This you can chnage if you want 
  xxi = linspace(min(xi),max(xi),n)' ;
  yyi = c(i)*ones(n,1) ; 
  zzi = interp1(xi,zi,xxi) ; 
  iwant{i} = [xxi yyi zzi] ; 
end
iwant = cell2mat(iwant) ; 

subplot(121)
scatter(x,y,50,z,'filled')
colorbar
title("Original")
subplot(122)
scatter(iwant(:,1),iwant(:,2),50,iwant(:,3),'filled')
colorbar
title("downsampled to best") ;