[Ffmpeg-devel] question about motion vectors

[Nguyen Trong Hung]

Dear all ffmpeg developers,

I am a final year student doing my thesis about video compression. One
of the main purposes of my thesis is to extract motion vectors during
the decoding process and store them to a seperate file. At first, I
thought that it would be difficult to achieve, when I browsed the
mailing list, I found that ffmpeg seems to already store motion
vectors in an array in case of DEBUG info is enabled.
I want to ask if my observation is true or not. If not then could
anyone point me some directions that i can go with ffmpeg.
My supervisor suggested me to work with MPEG2 first.

I worked with Nation ICT Australia (NICTA) and this is the base for
transform their algorithms of object recognision, tracking to
compressed domain. My interest is to extract motion vectors, rearrange
IBP frame, inverse DCT without fully decoding frames. I have just
started my thesis and really appreciate help and advices from ffmpeg
developers.

Thanks all for your time.

Best Regards