Segmentation fault

[GoogleCodeExporter]

trying to encode a bundle of tif images (other types also tried, but no success)

jbig2 -s -p -v -useOcr -autoThresh *.tif

I get this error:

---------------------------------------------------------
recognizing letters using OCR
Please call SetImage before attempting recognition.Please call SetImage
before attempting recognition.Segmentation fault
--------------------------------------------------------

I built jbig2enc modified with gcc 4.1.2 + leptonica 1.67 + tesseract 3.0 
headers

I attach also a sample tif file

Original issue reported on code.google.com by dria...@gmail.com on 12 Oct 2011 at 11:57

Attachments:

• 0373.tif

[GoogleCodeExporter]

I am using gcc 4.4.5 + leptonica 1.67 + tesseract 3.01 and it works fine, so I 
suggest try to use tesseract version 3.01.

I am attaching also output of jbig2 as one file without option -p (not in 
format suitable for putting in PDF, but it works with this option as well).

You can decode it for example using jbig2dec -o decodedImage.png out.jb

Original comment by hata.ra...@gmail.com on 12 Oct 2011 at 7:53

Attachments:

• out.jb

[GoogleCodeExporter]

Original comment by hata.ra...@gmail.com on 12 Oct 2011 at 7:55

• Changed state: Started

[GoogleCodeExporter]

thanks

I used this package

tesseract-3.00.tar.gz

downloaded from:

http://code.google.com/p/tesseract-ocr/downloads/detail?name=tesseract-3.00.tar.
gz&can=2&q=

I'll try with latest svn version (r 628, that I have just downloaded)

Original comment by dria...@gmail.com on 12 Oct 2011 at 8:39

[GoogleCodeExporter]

Great, just let know, if it helped or not.

Original comment by hata.ra...@gmail.com on 13 Oct 2011 at 10:01

[GoogleCodeExporter]

[deleted comment]

[GoogleCodeExporter]

I tried to build again (r88 of jbig2enc modified)

using same environment (leptonica 1.67 + tesseract 3.0 headers)

but, this time, I compiled with gcc 4.2.2

and I successful built finally your jbig2enc

I made some tests with and without -useOcr switch

0368.tif
jbig2 -s -p -v -useOcr -autoThresh
source image: 3577 x 4396 (1 bits) 600dpi x 600dpi, refcount = 1
thresholded image: 3577 x 4396 (1 bits) 600dpi x 600dpi, refcount = 2
JBIG2 compression complete. pages:1 symbols:1699 log2:11

jbig2 -s -p -v 
source image: 3577 x 4396 (1 bits) 600dpi x 600dpi, refcount = 1
thresholded image: 3577 x 4396 (1 bits) 600dpi x 600dpi, refcount = 2
JBIG2 compression complete. pages:1 symbols:1705 log2:11
------------

0-109.png
jbig2 -s -p -v -useOcr -autoThresh
source image: 2778 x 3881 (32 bits) 300dpi x 300dpi, refcount = 1
thresholded image: 2778 x 3881 (1 bits) 300dpi x 300dpi, refcount = 1
JBIG2 compression complete. pages:1 symbols:2779 log2:12

jbig2 -s -p -v
source image: 2778 x 3881 (32 bits) 300dpi x 300dpi, refcount = 1
thresholded image: 2778 x 3881 (1 bits) 300dpi x 300dpi, refcount = 1
JBIG2 compression complete. pages:1 symbols:2781 log2:12

Original comment by dria...@gmail.com on 14 Oct 2011 at 11:35

Attachments:

• 0368.tif
• 0-109.png

[GoogleCodeExporter]

Original comment by hata.ra...@gmail.com on 14 Oct 2011 at 11:43

• Changed state: Done

[GoogleCodeExporter]

that's great, don't forget, that now the integration with tesseract api 
is only in progress and right now there are only some outputs which 
doesn't modify the result. But I am currently working on it and I guess 
some beta version using tesseract should appear quite soon (probably 
during next month). But testing can take some time. If you have some 
idea for making some semi-automatic tests let me know.

Original comment by hata.ra...@gmail.com on 14 Oct 2011 at 11:46