Article View: comp.compression.research
Article #4Re: Compression versus Bit Rate Reduction?
From: Thomas Kinsman
Date: Fri, 30 Oct 1992 16:36
Date: Fri, 30 Oct 1992 16:36
48 lines
3190 bytes
3190 bytes
In article <...> madler@cco.caltech.edu (Mark Adler) writes:
> Actually, I have my own terms for those things that I would now like
> to propose for general adoption: I call coding of redundancy and
> entropy and any other sort of lossless compression "source-coding",
> since it depends on characteristics of the source.
> Selectively tossing out information for compression, for example
> visual or audio information not considered perceptible, I call
> "sink-coding", since it depends on characteristics of the
> eventual destination of the data, which can be called the sink.
These are semantics issues, but the subtle interpretations of the
terminology are important. Each term communicates different connotations.
The problem with using "source" and "sink" is that they are used in many
other ways, and can quickly confuse the issue. Especially when speaking
verbally, when "sink" and "synch" (as in synchronization) can become
confused. Remember, that if you are sending video information, part of
the channel includes "synch" pulses.
There is no set terminology. However, some terms you will find in the
literature which span the ideas you are talking about are:
A. LOSSLESS COMPRESSION
This communicates that no information is lost. Categories within
lossless compression describe the methods used to achieve lossless
compression.
B. LOSSY COMPRESSION.
Of course, this communicates that the information content is reduced.
Two other terms are used to describe the what Mark refers to as "source"
and "sink" coding:
i. REDUNDANCY CODING
This is implies entropy coding; some means of de-correlating the data.
ii. RELEVANCY CODING
This codes only the information which is relevant. So, if the Human
is the final destination, and the the information is below perceptible
levels it is not relevant. If you are transmitting radar data between
FAA flight controllers, the relevant information might be location,
altitude, and velocity.
For humans observers, the terms: "visually lossless" (images), "audibly
lossless" (audio), or "perceptually lossless" (both or either), connote
relevancy coding.
- Thomas
--
thomas@acadia.kodak.com Voice: 716/477-2831(w) Fax: 716/722-0160
Thomas B. Kinsman, Eastman Kodak Co., Flr 3, Bldg 65, RL, Roch., NY 14650-1816
Technical Details
Message-ID:
<1992Oct30.163614.6523@kodak.kodak.com>Path:
arc1-ca.pugleaf.net!sparky!uunet!charon.amdahl.com!pacbell.com!sgiblab!zaphod.mps.ohio-state.edu!rpi!newsserver.pixel.kodak.com!kodak!acadia!thomasReferences:
<BwuMs2.EnA@usenet.ucs.indiana.edu> <1cnjneINN7pr@gap.caltech.edu>Show Full Headers
Complete Article Headers
Newsgroups: comp.compression.research Path: arc1-ca.pugleaf.net!sparky!uunet!charon.amdahl.com!pacbell.com!sgiblab!zaphod.mps.ohio-state.edu!rpi!newsserver.pixel.kodak.com!kodak!acadia!thomas From: thomas@acadia.Kodak.COM (Thomas Kinsman) Subject: Re: Compression versus Bit Rate Reduction? Message-ID: <1992Oct30.163614.6523@kodak.kodak.com> Followup-To: comp.compression.research Summary: REDUNDANCY VS. RELEVANCY CODING Keywords: redundancy,relevancy,lossless,lossy Sender: THOMAS B. KINSMAN Organization: Eastman Kodak Co., Rochester, NY References: <BwuMs2.EnA@usenet.ucs.indiana.edu> <1cnjneINN7pr@gap.caltech.edu> Date: Fri, 30 Oct 92 16:36:14 GMT Lines: 48