Forum Jump

Select Group :
Select Forum :
Sorted By :
Sort Order :
From The :
 
Thread Tools  Search this thread 
vepro_hoffmann
Total Points:
170
Status Points:
0
Green Belt
July 1, 2009 7:32 AM PDT
JPEG lossless encoding problem
Hello,

I'm currently running into problems while encoding some 16bit grayscale pictures to JPEG lossless using code from the JPEGView exampe.
My code is working for most of the pictures but some just abort with JPEG_ERR_INTERNAL when returning from EncodeHuffmanMCURowLS.

Picture information:
witdh = 2320
height = -2920
bits per pixel  = 16 bit
number of planes = 1

Do you have any idea what is happening here? I think that every value should be valid as source pixel when encoding JPEG lossless.
I've attached the uncompressed pixeldata and a screenshot of the image.
 Attachments 
example.png
uncompressed.zip
Gennady Fedorov (Intel)
Total Points:
12,122
Status Points:
11,622
Brown Belt
July 3, 2009 7:51 AM PDT
Rate
 
#1
What IPP and ipp-sample versions you are using?
ipp_sample ( look at support.txt file you can find ther smth. like Package ID: w_ipp-samples_p_6.1.0.038)
--Gennady
vepro_hoffmann
Total Points:
170
Status Points:
0
Green Belt
July 6, 2009 12:20 AM PDT
Rate
 
#2 Reply to #1
What IPP and ipp-sample versions you are using?
ipp_sample ( look at support.txt file you can find ther smth. like Package ID: w_ipp-samples_p_6.1.0.038)
--Gennady


I was using IPP5.3 Update 4 and updated my code to w_ipp-uic_p_6.0.0.130 and w_ipp_ia32_p_6.0.2.074.
The problem stayed the same with the newer version.
vepro_hoffmann
Total Points:
170
Status Points:
0
Green Belt
July 13, 2009 2:05 AM PDT
Rate
 
#3 Reply to #2
Any news? I don't think that IPP6.1 would fix this issue.
I can trace the problem into the UIC code right to the ippiEncodeHuffman function. The input seems valid to me.. and as I said before I can't think of any invalid input for huffman encoding in general.
sryadno
Total Points:
170
Status Points:
120
Green Belt
July 14, 2009 12:41 AM PDT
Rate
 
|Best Answer
#4 Reply to #3
Quoting - vepro_hoffmann
Any news? I don't think that IPP6.1 would fix this issue.
I can trace the problem into the UIC code right to the ippiEncodeHuffman function. The input seems valid to me.. and as I said before I can't think of any invalid input for huffman encoding in general.

Hi, it seems some errors in our UIC jpeg codec in LOSSLESS mode. Try to do next fixes:

1) In uic_jpeg_enc.h file change
int GetPredictor(void) { return m_pntTransform; } to       int GetPredictor(void) { return m_predictor; }

2) in jpegenc.cpp change

buflen = (m_jpeg_mode == JPEG_LOSSLESS) ? IPP_MAX(ENC_DEFAULT_BUFLEN,m_numxMCU * m_jpeg_ncomp * 2) :     ENC_DEFAULT_BUFLEN;

to


buflen = (m_jpeg_mode == JPEG_LOSSLESS) ? IPP_MAX(ENC_DEFAULT_BUFLEN,m_numxMCU * m_jpeg_ncomp * 2 * 2) : ENC_DEFAULT_BUFLEN;
vepro_hoffmann
Total Points:
170
Status Points:
0
Green Belt
July 15, 2009 5:09 AM PDT
Rate
 
#5 Reply to #4
The second change did it. The buffersize was too small and that's why the encoding failed.
It's also the same problem as in IPP5.3 Update 4.

Thank you very much sryadno!


Intel Software Network Forums Statistics

8285 users have contributed to 31229 threads and 99106 posts to date.
In the past 24 hours, we have 17 new thread(s) 64 new posts(s), and 100 new user(s).

In the past 3 days, the most popular thread for everyone has been comparison cilk++, openmp, pthreads first results The most posts were made to comparison cilk++, openmp, pthreads first results The post with the most views is Unless I'm misreading this,

Please welcome our newest member tvinni