Save unique variable after same routine of n loops and output to a table or file?

I have a code that takes a .jpg file and performs some pixel analysis functions to it and outputs a floating variable. I want to save this unique variable after looping the routine for n .jpg files in a directory. Then I want to tabulate all of these variables and write them to a file.

Why does the below not allow me to do it?

jpegFiles = fullfile('Users/myfolder/Desktop/thresholding/*.jpg');
for k = 1:length(jpegFiles)

   filename = strcat(num2str(k), '.jpg');
   if exist(filename, 'file');
      fprintf('%s successfully read.\n', filename);
      a = imread(filename, 'jpg');  

      % Calculation functions omitted for brevity

      Cov = somefunction;
      fprintf('\n%s%.2f\n','Image pixel coverage:',Cov);

%why won't this let me store Cov variables as a matrix? It overrides the previous Covs if I try the below.

      for i = 1:k
          Cov2  = num2str(Cov); %create string from number         
          fprintf('The coverage as a string is:',Cov2);
      end 

      %This line forward prints only the last value to file in same dir as image files but I want it to do so for all Covs:

      fid = fopen('Output.csv','wt');
      fprintf(fid,'%0.4f\n',Cov);
      fclose(fid);

    end

Answer 1

[Walter Roberson]

You have

for k = 1:length(jpegFiles)

so your "k" is a scalar.

You have

for M = 1:length(k);

Because k is a scalar, length(k) is 1, so you are effectively using

for M = 1:1

The natural immediate reading would be to think that you wanted

for M = 1 : k

but if you do that then you are going to be writing the same unchanged COV matrix to every element of A, since COV does not change in you "for M" loop. This suggest that a "for M" loop is not desirable at that point in the code.

You have

dlmwrite('kerogen coverage.xls',A)

As per the above you only ever write to A{1}, so you will only ever have one output set in the file with your code. You are also always writing to the same file which would keep overwriting that one output set with a new value.

I suspect that what you want is to change

      for M = 1:length(k);
          A{M} =  Cov;       
          fprintf('\n%s%.2f\n','This value is stored in the percentage coverage matrix:',A{M});
          celldisp(A);
      end

to

          A{k} =  Cov;       
          fprintf('\n%s%.2f\n','This value is stored in the percentage coverage matrix:',A{k});
          celldisp(A);

In this way A is going to get larger on each k iteration, and will always include the previous output, so although you overwrite the same file each time, the file will get updated as the run progresses.

You could have a problem if the dlmwrite() fails as you could end up scribbling over everything you have done so far. That is why it is not a good idea to continually write over all of a file.

[Svet Shkolyar]

Success!! Thank you so much, Walter!!!

Now how do I get the output to be a table in my csv file that is tab separated rather than comma-separated?

[Walter Roberson]

dlmwrite('kerogen coverage.xls', A, '\t')

Answer 2

[Walter Roberson]

It does not allow you to do that as a deliberate design decision to reduce the number of times that people use this kind of bad programming technique. Dynamic variable names almost always cause problems and should be avoided whenever possible.

http://matlab.wikia.com/wiki/FAQ#How_can_I_create_variables_A1.2C_A2.2C....2CA10_in_a_loop.3F

[Svet Shkolyar]

Ok, thanks. I see. So I now changed the last part to the following. It seems to work better now but how do I just get it to display all of the Cov variables stored in A{i}?

fprintf('\n%s%.2f\n','Kerogen coverage:',Cov);

%Creating a matrix of Covs so all others don't get overwritten.

      for i = 1:length(k);
          A{i} =  Cov;       
          fprintf('\n%s%.2f\n','The values stored in the matrix are:',A{i});
      end

[Svet Shkolyar]

Update! I figured out to use

celldisp(A)

to display my cell's contents but the Cov variable is continuing to get overwritten each loop. Why isn't it storing the unique variable and putting into cell A after each loop?

[Walter Roberson]

We need to see your revised code

[Svet Shkolyar]

jpegFiles = fullfile('Users/myfolder/Desktop/thresholding/*.jpg');
for k = 1:length(jpegFiles)

   filename = strcat(num2str(k), '.jpg');
   if exist(filename, 'file');
      fprintf('%s successfully read.\n', filename);
      a = imread(filename, 'jpg');  

      filename1 = strcat(filename,'_flagged.mat');  

      red_min = 0;  
      red_max = 13; 
      green_min = 0; 
      green_max = 13;
      blue_min = 0;
      blue_max = 13;

      a = imread(filename);

      a_red = a(:,:,1);
      a_green = a(:,:,2);
      a_blue = a(:,:,3);
      ind_flag = find( a_red >= red_min & a_red <= red_max & a_green >= green_min & a_green <= green_max & a_blue >= blue_min & a_blue <= blue_max );
      flagged = zeros(size(a_red));
      flagged(ind_flag) = 1; 
      figure
      image(uint8(cat(3,255*ones(size(flagged)),255*(ones(size(flagged))-flagged),255*(ones(size(flagged))-flagged))))
      figure
      image(uint8(a))
      save(filename1,'flagged')  

      %STEP 2

      load (filename1)

      filename = strcat(filename1,'_particles.mat'); 

      flagged = padarray(flagged,[1,1],0); %Pad the marked pixel array to avoid a different neighbor index algorithm at the edges
      ind = find(flagged > 0);
      flagged(ind) = Inf;
      particles = zeros(size(flagged));
      neighbors = zeros(size(flagged));
      [m,n] = size(flagged);  
      neighbors_offset = [m,-m,1,-1,m+1,m-1,-m+1,-m-1];
      j = 1;

      while( min(size(find(isinf(flagged) == 1))) > 0 )
          particles(ind(1,1)) = j; %Save index location
          flagged(ind(1,1)) = 0; %Unmark pixel
          neighbor_indices = bsxfun(@plus,ind(1,1),neighbors_offset); 

          neighbors(neighbor_indices) = 1;
          ind_neighbors = find(isinf(neighbors.*flagged) > 0);
          neighbors(neighbor_indices) = 0;

          while( min(size(ind_neighbors)) > 0 ) %While you have neighbors to check
              particles(ind_neighbors) = j;
              flagged(ind_neighbors) = 0;
              neighbor_indices = bsxfun(@plus,ind_neighbors,neighbors_offset); 
              neighbors(neighbor_indices) = 1;
              ind_neighbors = find(isinf(neighbors.*flagged) > 0); 
              neighbors(neighbor_indices) = 0;
          end

          neighbors = 0*neighbors;
          j = j+1;  
          ind = find(flagged > 0); %Update the to-be-checked list
      end

      particles = particles(2:end-1,2:end-1); %Remove the padding
      save(filename,'particles');  
      figure
      pcolor(particles);
      shading interp

      load(filename)

      filename = strcat(filename1,'_particles_filtered.mat');

      particle_threshold = 2; %The minimum area, in pixels, to be considered a grain. Everything below this size is discarded. 

      num_particles_before = max(particles(:)); %Number of particles before filtering
      num_particles_after = 0; %Number of particles after filtering
      particles_after = zeros(size(particles)); %Storage for filtered particles

      for j = 1:num_particles_before
          ind = find(particles == j);

          if( max(size(ind)) > particle_threshold )
              num_particles_after = num_particles_after+1;
              particles_after(ind) = num_particles_after;
          end    
      end

      save(filename,'particles_after');
      ind = find(particles > 0);
      particles(ind) = 1;
      ind = find(particles_after > 0);
      particles_after(ind) = 1;
      figure
      image(uint8(cat(3,255*ones(size(particles)),255*ones(size(particles))-255*particles,255*ones(size(particles))-255*particles)));
      figure
      image(uint8(cat(3,255*ones(size(particles)),255*ones(size(particles))-255*particles_after,255*ones(size(particles))-255*particles_after)));
      figure
       image(a)

      load (filename1); %Generated from step3.m

      num_particles = max(particles_after(:)); %Total number of particles

      size_of_particles = zeros(1,num_particles); %Size of each, in pixels
      for j = 1:num_particles
          ind = find(particles_after == j);
          size_of_particles(1,j) = max(size(ind));
      end
      average_size = mean(size_of_particles);
      var_of_size = var(size_of_particles);
      percentage_of_slide = sum(size_of_particles)/prod(size(particles_after));
      Cov = 100 - 100*percentage_of_slide;

      fprintf('\n%s%i%s\n','There are ',num_particles,' total particles')
      fprintf('\n%s%.2f%s\n','The average size is ',average_size,' pixels')
      fprintf('\n%s%.2f%s%.2f\n','The size has variance ',var_of_size,' and standard deviation ',sqrt(var_of_size))
      fprintf('\n%s%.2f\n','Kerogen coverage:',Cov);

%Creating a matrix of Covs but they keep getting overwritten!

      for M = 1:length(k);
          A{M} =  Cov;       
          fprintf('\n%s%.2f\n','This value is stored in the percentage coverage matrix:',A{M});
          celldisp(A);
      end 

      dlmwrite('kerogen coverage.xls',A)

    else   

          % No else statement needed.

    end  
  end