write multiple txt files

hello, i want to read 45 images. these images are binarized and their names is numbered from 1 to 45 and then count number of black points "zeros" in each image and set the result in txt file with the same name of the image but i have an error in the line of " I=imread( H '.tif' ).

i think that i have syntax error but i dont know what is it

for H=1:45
I=imread( H '.tif');
J = I;
FileName=[ H '.txt'];
file=fopen(FileName,'wt');
[rows,cols]=size(J);

 fprintf('Rotated: %6.2f.\n');
for r=1:rows
    ct=0 ;
    for c=1:cols
        pic=J(r,c);
        if pic == 0
           ct = ct+1 ;
        else
            %%gray(i,j)=0;
        end
        
               
    end
    fprintf(file,'%2.0f, \n',ct);
    fprintf('Rotated: %6.2f.\n',ct);
end
fclose(file);
end

Answer 1

[Rik]

A few lines later you try something better when you try to form a file name. But the problem is a bit deeper: you are trying to implicitly convert a number to its equivalent char. You should do that explicitly.

The code below still has some issues that you need to address.

for H=1:45
    I=imread(sprintf('%d.tif',H));%consider using a full path
    J = I;
    FileName=sprintf('%d.txt',H);
    file=fopen(FileName,'wt');
    [rows,cols]=size(J);
    fprintf('Rotated: %6.2f.\n');%no input, and printed to the command line
    for r=1:rows
        ct=0 ;
        for c=1:cols
            pic=J(r,c);
            if pic == 0
                ct = ct+1 ;
            else
                %%gray(i,j)=0;
            end
            
            
        end
        fprintf(file,'%2.0f, \n',ct);
        fprintf('Rotated: %6.2f.\n',ct);%printed to the command line
    end
    fclose(file);
end

You could also use the sum function instead of this loop:

for H=1:45
    I=imread(sprintf('%d.tif',H));%consider using a full path
    J = I;
    FileName=sprintf('%d.txt',H);
    file=fopen(FileName,'wt');
    [rows,cols]=size(J);
    fprintf('Rotated: %6.2f.\n');%no input, and printed to the command line
    sums_list=sum(J==0,2);
    for r=1:rows
        ct=sums_list(r);
        fprintf(file,'%2.0f, \n',ct);
        fprintf('Rotated: %6.2f.\n',ct);%printed to the command line
    end
    fclose(file);
end

[Mahmoud Hassan]

first, now i face these errors ... and i dont need to sum because i want the dark points in each row so i just need to count then ct = 0 after each row

note: i am using matlab 2014

[Mahmoud Hassan]

do you have any idea why i get these errors ?!

[Rik]

The imread error is something you need to solve from your end. The creation of the file name goes as it should:

>> sprintf('%d.tif',4)
ans =
4.tif

And did you actually look at what I did with the sum? I didn't take the sum of the pixel values, but I first converted it to a logical matrix where all pixels with intensity 0 would be true. This means you can skip the col-loop. You could also omit the row-loop:

for H=1:45
    I=imread(sprintf('%d.tif',H));%consider using a full path
    file=fopen(sprintf('%d.txt',H),'wt');
    fprintf('Rotated: %6.2f.\n');%no input, and printed to the command line
    sums_list=sum(I==0,2);
    fprintf(file,'%2.0f, \n',sums_list);
    fprintf('Rotated: %6.2f.\n',ct);%printed to the command line
    fclose(file);
end

This code is all predicated on the assumption that you actually can read the image file and that it only contains intensity values, and not RGB-values.

[Mahmoud Hassan]

I have understood what you said and your code ... but i dont know why i get the same damn error :/

[Rik]

It appears that Matlab has trouble determining the format. Maybe Matlab expects tiff-files to have the extension tiff instead of tif.

To remedy this, the documentation (R2014b) explains the syntax for forcing the format:

A = imread(sprintf('%d.tif',H), 'tiff')

Also, I have just noticed I made a mistake in copy-paste. If you look at the .txt file, I forgot to change the extension to .txt, which caused the images to be overwritten. I'll edit my posts to fix this.

[Mahmoud Hassan]

it worked bro ! really thank you so much ... and yeah i noticed it that wasnt the problem ... but i have one more question pls ... does your code would have different result for number of dark points in each row from my code ? i mean is my logic to calculate the dark points is wrong or less effective than your sums_list ?!

[Rik]

It should be the exact equivalent output, but my code should get there much faster.

With increasing size the benefit grows. This also has to do with the expected number of zero values in your image, see the block below.

The sum method takes 29.5 % of the time that the loop takes (@ size(I)=[1000 1], total time= 0.004 seconds)
The sum method takes 20.7 % of the time that the loop takes (@ size(I)=[1000 10], total time= 0.005 seconds)
The sum method takes 15.8 % of the time that the loop takes (@ size(I)=[1000 100], total time= 0.027 seconds)
The sum method takes 8.6 % of the time that the loop takes (@ size(I)=[1000 1000], total time= 0.159 seconds)
The sum method takes 5.9 % of the time that the loop takes (@ size(I)=[1000 10000], total time= 3.326 seconds)
The sum method takes 3.7 % of the time that the loop takes (@ size(I)=[1000 100000], total time= 49.634 seconds)

This was generated with this code:

clc
%5 takes about 3.5 seconds on my machine, 6 already 50
timing_perc=zeros(1,6);
repeats=20;%keep high to account for JIT optimization
for sz=0:(numel(timing_perc)-1)
    I=randi([0 10],1000,10^sz);
    
    out1=zeros(size(I,1),1);
    out2=out1;
    tic
    for n=1:repeats
        out1=sum(I==0,2);
    end
    t1=toc;
    tic
    for n=1:repeats
        [rows,cols]=size(I);
        for r=1:rows
            ct=0 ;
            for c=1:cols
                pic=I(r,c);
                if pic == 0
                    ct = ct+1 ;
                else
                    %%gray(i,j)=0;
                end
            end
            out2(r)=ct;
        end
    end
    t2=toc;
    %fprintf('Elapsed time is %8f seconds for sum.\n',t1)
    %fprintf('Elapsed time is %8f seconds for loop.\n',t2)
    fprintf(['The sum method takes %.1f %% of the time that the ',...
        'loop takes (@ size(I)=[%d %d], total time= %.3f seconds)\n'],...
        100*t1/t2,size(I,1),size(I,2),t1+t2)
    timing_perc(sz+1)=100*t1/t2;
end

figure(1),clf(1)
plot(0:sz,100-timing_perc)
xlabel('order of magnitude of image size')
ylabel('percentage of time gain')
ylim([50 100])