I'm not an expert on video codecs either, but that is yet another math patent.
It includes a reference to an "apparatus" in the summary, and a ref to "storage means" [sic] (should be "storage medium") in one of the claims, but that claim (4) is the only one I noticed that makes any reference to any non-theoretical, non-mathematical, non-ephemeral (i.e. the vague notion of "transmitting" or "retrieving") element.
If patenting math weren't bad enough, completely absurd chained claims include:
3. A method according to claim 2 further comprising the steps of: transmitting said coded information to a decoder for decoding.
4. A method according to claim 2 further comprising the steps of: storing said coded information in a storage means.
42. A video encoder according to claim 41, further comprising: means for transmitting said encoded information for the segment to a decoder for decoding.
If claims 2 and 41 (base claims for 3, 4, and 42) are too weak to apply in a particular case, or are unpatentable (which should be the case since they are math claims), they cannot possibly be made stronger simply by handwaving in some mentions of storage and of transferring data. Yet that is how all of these algorithm/math patents hope to survive. Reference a computer, referencing storing something, reference transmitting something, and suddenly something that's not patentable becomes patentable.
This patent racket depends on the idea that it's so uncommon to think to use a computer to perform some calculation, or implement an algorithm in a chip, or use disks or memory to store data, or use a data bus or ethernet/infiniband/rapidio to transmit anything, that adding those things (usually not even mentioning a specific type of computer, or storage, or transmission mechanism) vaults the idea into uniqueness.
Supposing 3, 4, and 42 somehow survived, because they might in the strictest sense be considered to involved material things, they amount to retrieving data (from an unspecified data source), performing transformations (math) to them (using unspecified hardware), and possibly storing or transmitting the results (again, hardware unspecified).
Software patents are simply a scam. They take something that's unpatentable subject matter, add generic boilerplate to it, trying to connect it with real hardware, and hope the USPTO approves it.