montnaro%chenengo.tcpip@GE-CRD.ARPA (05/21/86)
Received: by chenengo.steinmetz (3.0/1.1x Steinmetz) id AA16370; Wed, 21 May 86 10:08:20 EDT Date: Wed, 21 May 86 10:08:20 EDT From: Skip Montanaro <montnaro%chenengo.tcpip@csbvax> Posted-Date: Wed, 21 May 86 10:08:20 EDT Message-Id: <8605211408.AA16370@chenengo.steinmetz> To: info-atari16@score.stanford.edu Subject: image compression In <752@rclex.UUCP> Walt Weber writes: >Has anyone ever seen or used an encoding or compression algorithm >to store image files? I have seen a couple of demos which appear >to be compressed, as they are all different sizes and come up in >a "slideshow" style presentation. I would think that some adaptation of the algorithm used in compress (available from mod.sources and quite protable) might be sufficiently fast enough to do what you want. Compress is not restricted to ASCII files. We use it to compress cpio output (used for backups) before sending to a VMS VAX. We get (typically) about 50% compression and its speed is acceptably fast. ========== Skip Montanaro ARPA: montanaro@ge-crd.arpa US Mail: General Electric Company Corporate Research and Development P.O. Box 8 Bldg KW, Room C210 Schenectady, NY 12304 Phone: 518-387-7312 ==========
jhs@disunix.UUCP (05/22/86)
A lot of study has gone into image compression, so I am sure you can find some very effective algorithms. One of the simplest to understand is "run-length coding", in which repeated data (pixel) values are sent as a special "repeat code" followed by the repeat count and the value, or some such thing (invent your own protocol if you like). For certain kinds of images, this results is a substantial compression. E.g. line drawings where most space is white anyway, or highly structured scenes where large regions have the same color and little or no texture. If you are constructing the image (i.e. have your mits on the semantics of the scene) then you can do MUCH better with "Videotex" encoding. The North American Presentation-Level Protocol (NAPLPS) does a surprisingly good job of representing simple pictures etc. There is an ANSII standards group (X3H33?) working on such things. With Videotex encoding (NAPLPS, Teletel, Prestel, several other standards), pictures can be coded in a few thousand bytes and therefore sent over even 1200-baud lines in a few seconds! That's about all I know except I could probably get some specific references to articles, standards documents, etc. if you can't scrounge them up locally. -John Sangster jhs@mitre-bedford.arpa