MATLAB Answers

0

How can I save data from features files into matrix?

Asked by bahaa rabi on 14 Oct 2019
Latest activity Commented on by per isakson
on 16 Oct 2019
i want to save data from 500 images features files every file contain 6 types of features types such as jcd,tamura,..... .,
the problem is that the data for these 6 features are seperated by comma,i want to make six matrices for these 500 image for type of features ,i atteched this file to help
can any one help

  2 Comments

"the problem is that the data for these 6 features are seperated by comma" Why is that a problem?
i mean that i want code that save the data for each feature for 500 images .so finally i get 6 matrices the size for each one 500*feature size data .like tamura the size will be 500*18.that i mean.the code will help me to save time to import data for 6 features for 500 images features.with thanks

Sign in to comment.

1 Answer

Answer by per isakson
on 15 Oct 2019
Edited by per isakson
on 15 Oct 2019
 Accepted Answer

I understand that "data for each feature for 500 images" is in ONE text file with 500 blocks of six rows. The format is illustrated by im1.txt which contains two blocks.
Try this
>> S = cssm('im1.txt')
S =
1×2 struct array with fields:
JCD
Tamura
ColorLayout
EdgeHistogram
AutoColorCorrelogram
PHOG
>> JCD_matrix = cat( 1, S(:).JCD );
>> whos JCD_matrix
Name Size Bytes Class Attributes
JCD_matrix 2x168 2688 double
where (in one m-file)
function im = cssm( ffs )
%%
fid = fopen( ffs, 'rt' );
buf = textscan( fid, '%s', 'Delimiter', '\n' );
fclose( fid );
buf = buf{1};
%%
len = floor( size(buf,1)/6 ); % number of blocks
im(1,len) = struct( 'JCD',[], 'Tamura',[], 'ColorLayout',[] ...
, 'EdgeHistogram',[], 'AutoColorCorrelogram',[], 'PHOG',[] );
%%
for jj = 1 : len
im(jj).JCD = read( buf{(jj-1)*6 + 1 }, 'JCD' , (jj-1)*6+1 );
im(jj).Tamura = read( buf{(jj-1)*6 + 2 }, 'Tamura' , (jj-1)*6+2 );
im(jj).ColorLayout = read( buf{(jj-1)*6 + 3 }, 'ColorLayout' , (jj-1)*6+3 );
im(jj).EdgeHistogram = read( buf{(jj-1)*6 + 4 }, 'EdgeHistogram' , (jj-1)*6+4 );
im(jj).AutoColorCorrelogram = read( buf{(jj-1)*6 + 5 }, 'AutoColorCorrelogram', (jj-1)*6+5 );
im(jj).PHOG = read( buf{(jj-1)*6 + 6 }, 'PHOG' , (jj-1)*6+6 );
end
end
function row = read( chr, name, n )
cac = strsplit( chr, ':' );
assert( strcmp(cac{1},name),'Row, %d, is "%s". Expected is "%s"', n, cac{1}, name )
num = textscan( cac{2}, '%f', 'Delimiter',',' );
row = reshape( num{1}, 1,[] );
end
If execution time is a problem replace
cac = strsplit( chr, ':' );
by
cac = regexp( chr, ':', 'split' );

  5 Comments

I cannot see that "Error: File: Untitled3.m Line: 1 Column: 1" could be caused by my code.
"Undefined function or variable 'cssm'" is because you have not copied and pasted the function I provide in my answer to an m-file, i.e the text that follows after "where (in one m-file)". See function Declare function name, inputs, and outputs
very thanks i tried again this 2 codes and got results for only one im . what i do call the all 6 features for 500 im. my images strates from im1 to im5 , and i want 6 feature matrix like for matrix for jcd contains the jcd features for 500 images. thank you for your cooperation.
I wrote: "I understand that "data for each feature for 500 images" is in ONE text file with 500 blocks of six rows. The format is illustrated by im1.txt which contains two blocks."
Why don't you say that I missinterpreted the question?

Sign in to comment.