Why are blocking artifacts serious when there is fast motion in MPEG?

Here is the guess I made:

In MPEG, each block in an encoding frame is matched with a block in the reference frame. If the difference of two blocks is small, only the difference is encoded using DCT. Is the reason blocking artifacts are serious that the difference of two blocks is too large and DCT cut the AC component?

  • 1
    $\begingroup$ This question may be offtopic; it seems to ask for a problem of a specific codec. If you want to make it better fit for this site, please add what you mean by "serious" and what "DCT" is. $\endgroup$ – Raphael Apr 22 '12 at 12:02
  • 2
    $\begingroup$ @Raphael Questions don't have to be worded for the least common denominator. There's nothing wrong with a question about the operation of a specific compression function. $\endgroup$ – Gilles 'SO- stop being evil' Apr 22 '12 at 17:23

DCT is both discrete and finite, it has a limited range in both the space and frequency domains. If the change in a certain location is large or the location changes a lot, the results of the transform will be beyond the sampling range in the frequency domain.

Since as you mentioned MPEG is based on blocks, the result of a block with with missing high frequencies, after transform back to the spacial domain, will result in the loss of detail in that block - making the block look flat.

For a more accurate result try Signal Processing Stack Exchange.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.