Print OCR results to text file

Hello,

I have read the other posts and examples for printing your OCR results into a text file but am running into an error.

Error using fprintf
Function is not defined for 'cell' inputs.

Error in REMOVEDallEXTRAbits (line 142)
fprintf(fileID, '%s\r\n', text);
 

ocrtxt = ocr(BW2, textBBoxes, 'Language', 'C:\Users\e362609\Documents\MATLAB\Examples\vision\TextDetectionExample\try50\tessdata\try50.traineddata');
% [ocrtxt.Text];

fileID = fopen('trytext.txt');
fprintf(fileID, '%s\r\n', text);
fclose(fileID);

Any help is much apreciated!

[Star Strider]

I don’t have the Computer Vision Toolbox, so I have no personal experience with it.

However (according to the online documentation on ocr), it appears that the information you want is either in the structure fields:

ocrtext.Text

that appears to return a character array, or:

ocrtext.Words 

that returns a cell array.

Consider that:

ocrtxt.Words = {'This','is','a','test'};       % Create Something Similar To What I Believe ‘ocr’ Returns
fprintf(f,'%s\n',ocrtxt.Words{:})

produces:

This
is
a
test

and no loop is necessary.

See if that works with your data.

[Laura Elder]

f = fopen('C:\Users\e362609\Documents\MATLAB\Examples\vision\TextDetectionExample\trytext.txt', 'w+');
fprintf(f, '%s\r\n',ocrtxt.Words{:})
fclose('all');


Expected one output from a curly brace or dot indexing expression, but there were 544 results.

Error in REMOVEDallEXTRAbits (line 144)
fprintf(f, '%s\r\n',ocrtxt.Words{:})
 

Answer 1

[Laura Elder]

THE GOLDEN TICKET!!!!

THANK YOU ALL FOR ALL OF YOUR HELP :)

ocrtxt = ocr(BW2, textBBoxes, 'Language', 'C:\Users\e362609\Documents\MATLAB\Examples\vision\TextDetectionExample\try50\tessdata\try50.traineddata');
% ocrtxt(1,1).Text

f = fopen('C:\Users\e362609\Documents\MATLAB\Examples\vision\TextDetectionExample\trytext.txt', 'w+');
for i= 1 :numel(ocrtxt)
  fprintf(f,'%s\r\n',ocrtxt(i).Text);
end
% fclose('all');

Answer 2

[dpb]

To write cell output with fprintf, you must dereference the cell content "use the curlies, Luke!".

fprintf(fileID, '%s\r\n', text{:});

or, use the new writecell if you have R2019a.

[Laura Elder]

Hey dpb,

Thank you for the help.

I am now getting:

Undefined variable "text" or class "text".

Error in REMOVEDallEXTRAbits (line 144)
fprintf(fileID, '%s\r\n', text{:});
 

I have tried:

making the variable as:

text = [orctext.Words]

and got the below error

Error using horzcat
dimensions of arrays being concatenated are not consistant.

Error in ...
    text = [orctext.Words];

I have tried the above with orctext.Text and orctext.Words

[Laura Elder]

Error
Brace indexing is not supported for variables of
this type.

Error in REMOVEDallEXTRAbits (line 144)
fprintf(fileID, '%s\r\n', text{:});

and I am using MatLab R2018b

Expected one output from a curly brace or dot indexing expression, but there were 544 results.

Error in REMOVEDallEXTRAbits (line 142)
fprintf(fileID, '%s\r\n', ocrtxt.Text{:});

Error using fprintf
Invalid file identifier. Use fopen to generate a valid file identifier.

Error in REMOVEDallEXTRAbits (line 142)
fprintf(fileID, '%s\r\n', ocrtxt.Text);

[dpb]

I just used text as the variable name as that was the variable in the code you posted...

I have no idea what type an object or struct or what ocr returns; what does the doc say? I don't have whatever toolbox it is in...

BUT, the error message said "fprintf isn't defined for cell inputs" (which is true) and therefore I presumed you had passed a cell and the curlies {} are how to dereference a cell. So, something is different from what generated the previous error...

Start with

whos orctext

and show us what that returns.

[Laura Elder]

Here is an example of the types of images I am processing:

Here is the line that creates and defines ocrtext

ocrtxt = ocr(BW2, textBBoxes, 'Language', 'C:\Users\e362609\Documents\MATLAB\Examples\vision\TextDetectionExample\try50\tessdata\try50.traineddata');

Here is the output from running

whos orctext

It returned nothing

Here is ocrtext in the workspace

[dpb]

whos orctext

Misspelled the variable...

whos orctxt

[Laura Elder]

sorry

>> whos ocrtxt
  Name          Size              Bytes  Class      Attributes

  ocrtxt      544x1             1918916  ocrText              

>> 

Answer 3

[dpb]

AHA! The one about "Expected one output from a curly brace or dot indexing expression, but there were 544 results." is helpful! ocrtxt must be a struct and the Text field a cellstr array.

When we dereferenced with {:}, that gets the whole array as a comma-separated list--there apparently are actually 544 lines in the text you got back.

You need to output each cellstr...

for i=numel(ocr.Text)
  fprintf(fid,'%s\n',ocr.Text{i})
end

should provide a pathway to nirvana...presuming, of course, you've opened fid for write operations first...

ADDENDUM

The whos line showed an array of the object and so it isn't a single object with array of strings as the above presumes but an array with one string per element. So iterate over the array of objects instead of a (presumed) cellstr array..

for i=numel(ocrtxt)
  fprintf(fid,'%s\n',ocrtxt(i).Text)
end

If the doc is correct, this should NOT cause any issues about cell element in fprintf

[Laura Elder]

ocrtxt = ocr(BW2, textBBoxes, 'Language', 'C:\Users\e362609\Documents\MATLAB\Examples\vision\TextDetectionExample\try50\tessdata\try50.traineddata');

fileID = fopen('trytext2.txt');
for i=numel(ocrtxt.Text)
    fprintf(fileID, '%s\n', ocrtxt.Text{i});
end
fclose(fileID);

Expected one output from a curly brace or dot indexing expression, but there were 544 results.

Error in REMOVEDallEXTRAbits (line 144)
    fprintf(fileID, '%s\n', ocrtxt.Text{i});

Hey,

I am getting the same error. :(

I also tried

    ocrtxt = ocr(BW2, textBBoxes, 'Language', 'C:\Users\e362609\Documents\MATLAB\Examples\vision\TextDetectionExample\try50\tessdata\try50.traineddata');

fileID = fopen('trytext2.txt');
for i=numel(ocrtxt.Text)
    fprintf(fileID, '%s\n', ocrtxt.Text{i})
end
fclose(fileID);

Expected one output from a curly brace or dot indexing expression, but there were 544 results.

Error in REMOVEDallEXTRAbits (line 144)
    fprintf(fileID, '%s\n', ocrtxt.Text{i})

[Laura Elder]

I think that it would work if we were able to step up the second variable in incurmants of one.

So,

ocrtxt(1,1).Text

ocrtxt(1,2).Text

ocrtxt(1,3).Text

and append the txt file.

but this is not something that I know how to do in MatLab.

I'm an really new to codeing...

[Laura Elder]

ocrtxt = ocr(BW2, textBBoxes, 'Language', 'C:\Users\e362609\Documents\MATLAB\Examples\vision\TextDetectionExample\try50\tessdata\try50.traineddata');

fileID = fopen('trytext2.txt');
for i=numel(ocrtxt.Text)
    fprintf(fileID, '%s\n', ocrtxt(1,{i}).Text)
end
fclose(fileID);

It didn't work :(

Unable to use a value of type 'cell' as an index.

Error in REMOVEDallEXTRAbits (line 144)
    fprintf(fileID, '%s\n', ocrtxt(1,{i}).Text)

[dpb]

That's not the syntax I used...look at the sample code again.

But, the thing is an ocrtext object so need to look at the documentation to find out for sure how it is defined. The whos statement shows you have an array of 544 of 'em, though so will need to iterate thru the array...

What do

ocrtxt(1)
ocrtxt{1}

return? I'm presuming the second will fail; the first should give some klews about what it contains.

ADDENDUM

Unfortunately, the documentation is less than helpful on the subject... :(

Text — Text recognized by OCR
array of characters

Text recognized by OCR, specified as an array of characters. The text includes 
white space and new line characters.

.Text is a field of char it says, though, so you shouldn't run in to the cell problem...

Since it returns an array of the object class and not, as I had presumed, an object with a cellstr array, then the syntax is probably

for i=numel(ocrtxt)
  fprintf(fid,'%s\n',ocrtxt(i).Text)
end

to index the object array and return the .Text field within.

[Laura Elder]

hummm

yours:

for i=numel(ocr.Text)
  fprintf(fid,'%s\n',ocr.Text{i})
end

mine:

fileID = fopen('trytext2.txt');
for i=numel(ocrtxt.Text)
    fprintf(fileID, '%s\n', ocrtxt.Text{i})
end
fclose(fileID);

I also tried:

fileID = fopen('trytext2.txt');
for i=numel(ocrtxt.Text)
    fprintf(fileID, '%s\n', ocrtxt.Text{i});
end
fclose(fileID);

they gave me the same error as before

and

for i=numel(ocr.Text)
  fprintf(fid,'%s\n',ocr.Text{i})
end

tells me that ocr is not a defined variable

sorry if I am missing something obvious

[Laura Elder]

ok

I'll run

What do

ocrtxt(1)

ocrtxt{1}

return? I'm presuming the second will fail; the first should give some klews about what it contains.

It will take a few min.

one sec

[Laura Elder]

ocrtxt = ocr(BW2, textBBoxes, 'Language', 'C:\Users\e362609\Documents\MATLAB\Examples\vision\TextDetectionExample\try50\tessdata\try50.traineddata');
% ocrtxt(1,1).Text

f = fopen('trytext2.txt');
for i=numel(ocrtxt.Text)
  fprintf(f,'%s\n',ocrtxt.Text{i})
end
fclose(f);

Expected one output from a curly brace or dot indexing expression, but there were 544 results.

Error in REMOVEDallEXTRAbits (line 145)
  fprintf(f,'%s\n',ocrtxt.Text{i})

[Laura Elder]

>> ocrtxt(1)

ans =

ocrText with properties:

Text: '1D10↵↵'

CharacterBoundingBoxes: [6×4 double]

CharacterConfidences: [6×1 single]

Words: {'1D10'}

WordBoundingBoxes: [6932 8805 154 49]

WordConfidences: 0.9324

>>

>> ocrtxt{1}

Brace indexing is not supported for variables of this type.

>>

>> ocrtxt(1,1).Text

ans =

'1D10

'

>>

[Laura Elder]

f = fopen('trytext2.txt');
for i=numel(ocrtxt.Text)
  fprintf(f,'%s\n',ocrtxt(i).Text)
end
fclose(f);

Error
Array indices must be positive integers or logical values.

Error in REMOVEDallEXTRAbits (line 145)
  fprintf(f,'%s\n',ocrtxt(i).Text)
 

I double checked and I do not use i anyehere else in the file

I went though the help docs and someone had said that you might get this if you are trying to start at 0 and not at 1. So I tried :

  for i= 1 :numel(ocrtxt.Text)
  fprintf(f,'%s\n',ocrtxt(i).Text)
  end
  
  Error using fclose
Invalid file identifier. Use fopen to generate a valid file identifier.

Error in REMOVEDallEXTRAbits (line 147)
fclose(f);