How can I find nearest value in large grid data?

I have a grid having cells 100x60, containing contours data. Some of the cells have filled with whole number from 1 to 15 showing the elevation of contours, now how can I interpolate points between the contours line to fill all empty cells?

Note: Points are not in regular way, these scattered points

Answer 1

[Walter Roberson]

See John D'Errico's File Exchange contribution "inpaint_nan"

[Zubair Nizamani]

Dear Roberson thank you for your response and guideline. The function you have mentioned above is right now harder for me to understand because I am getting started with MATlab, So can you direct me any simplest way possible?

[Chad Greene]

If you download inpaint_nans, it makes the problem pretty easy. For example, say you have some gridded dataset like this:

Z = peaks(100);

but a few points are NaNs:

Z([423 898 1234] = NaN;

You can fill in the missing values like this:

Z_filled = inpaint_nans(Z);

Answer 2

[Image Analyst]

How about scatteredInterpolant()?

[Zubair Nizamani]

@Image Analyst, I have used the code mentioned below for interpolation, but I made an excel sheet in a way that, divided all the grid in 5x5 square boxes and then I assigned values on every corner of that boxes according to my best assumption, but when I plotted the contours, my contours lines were not smoothed and kinda uneven. So for now I want to make code that reads the excel as it and then interpolate.

If it is possible please do me favor here via helping me out.

[In excel sheet, only contour digitized values are present]

[Code]

clear all;

clc;

data = xlsread('Trial1.xlsx');

rlt = data ;

[height, width] = size(data);

[x,y] = meshgrid(0:5:width,0:5:height);

x(:,1) = 1; y(1,:) = 1;

z(1,1) = data(1,1);

for i = 1:height/5

      z(i+1, 1) = data(5*i, 1);
  end
  for j = 1:width/5
      z(1, j+1) = data(1, 5*j);
  end
  for i = 1:height/5
      for j = 1:width/5
          z(i+1,j+1) = data(5*i,5*j);
      end
  end
  [x1,y1] = meshgrid(1:width, 1:height);
  z1 = interp2(x, y, z, x1, y1, 'spline');
  for i = 1:height
      for j = 1:width
          if ~(data(height-i+1,j) > 0 )
              rlt(i,j) = z1(height-i+1,j);
     else
              rlt(i,j) = data(height-i+1,j);
          end
      end
  end
  % figure; mesh(x1,y1,z1); xlabel('longitude'); ylabel('latitude');
  % F = [.05 .1 .05; .1 .4 .1; .05 .1 .05];
  % ZC = conv2(rlt,F,'same');
  xlswrite('digital_value.xlsx', rlt);
  figure;
  %contour(rlt,20);
  [C,h] = contour(rlt);
  clabel(C,h);
  set(h, 'LevelList', [1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19], ...
         'TextList', [1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19])

[Image Analyst]

You forgot to attach the workbook. For the benefit of everyone, attach the workbook AND a screenshot of your plot or image.

What kind of help do you need reading Excel? You obviously know about xlsread(). Do you want to know about ActiveX also?

[Zubair Nizamani]

For instance, If this is my excel sheet or grid data,

0 0 0 1 0 0 2

0 0 1 0 0 2 0

0 1 0 0 2 0 0

1 0 0 2 0 0 0

0 0 2 0 0 0 0

Now can you tell me what is the best way to interpolate the point where zeroes are? and replace all the zeroes with interpolated data.

[Image Analyst]

I need to leave now, but my first attempt would be to use scatteredInterpolant, like I already suggested. But I'm sure you've tried that by now, so let's see your code - it may speed me up when I return later today. If you didn't try my suggestion, then why not?

[Walter Roberson]

YourData(YourData == 0) = nan;

Now call the File Exchange contribution inpaint_nan on YourData.