Ya - it's nowhere near "full, clear, concise and exact", that's why we're wallowing and wondering and wandering and weaving and ... Fortunately I enjoy figuring out puzzles like this and I have some history with image compression so the details don't scare me.
Thanks for the patent basis lookup brite_eye - I really would expect that this one doesn't get granted based on what's there. I'm not sure where they go from that - can you clean up your application and refile quickly? I'm sorta surprised considering the number of patents that these guys are tied to - seems sloppy.
Beyond being incomplete in spots (i.e. multiplex mechanism), there are things which are internally inconsistent. Lately I've been trying to make the histogram algorithms work...  says that all the histogram bins are initialized to 0, but then  explains that each bin must be at least 1 for arithmetic compression to work.
Then there is the claims section... Claims 20 and 22 are the same. Claim 23 seems like a rewrite of claim 21. Claim 12 looks like a combination of minor details in claims 10 and 11. WTF? Did they submit the rough draft?
Anyways... I'm trying to find a way through this mess to figure it out. There's also the possibility that the compression isn't what is described in this patent application - but where things are clearly described, the compression does appear to match the patent app. (So, we'll keep wandering through the maze looking for the way out.)