Path: utzoo!utgpu!water!watmath!clyde!att!osu-cis!killer!vector!telecom-request From: eli@spdcc.COM (Steve Elias) Newsgroups: comp.dcom.telecom Subject: CCITT facsimile standards Message-ID: Date: 27 Aug 88 13:18:49 GMT Sender: chip@vector.UUCP Lines: 28 Approved: telecom-request@vector.uucp (USENET Telecom Moderator) X-TELECOM-Digest: volume 8, issue 135, message 4 X-Submissions-To: telecom@xx.lcs.mit.edu (Mailing List Coordinator) X-Administrivia-To: telecom-request@vector.uucp (USENET Telecom Moderator) >From: robert@eva.slu.se (Robert Olsson datoravd, EMC tfn 018-172581) >Subject: image compressing in facsimile machines - algorithm wanted >Has anybody made an implementation of the codes in the CCITT T.4 telefax >recommendation. I think it's called modified Huffman code and is used to >compress scanned images in facsimile machines. I've stared at codes for >some time now and may be capable of encoding from run lengths to Huffman >codes - but decoding Huffman codes into run lengths ? To do that I would >definitely need the algorithm. Where can I find it ? Or code? yes. the coding scheme is as you describe. the code that i wrote for this is in one of the PC-Fax hardware/software products available today. it's not mine to give away, but i could probably give you yet more free advice, Robert, if you can get mail to me. the algorithm itself isn't that complex -- optimizing it is a different story. regarding the algorithm: the principle of Huffman coding allows you to determine the value of a bit string of any length -- not a single one of the code strings is the same as the prefix to any of the longer code strings. another factor: be sure to re-re-check the piles of Huffman and run length codes after you type them in. a single typo will result in nasty bugs. by the way, there should be an addendum to the CCITT fax spec soon, which will standardize gray-scale fax images as well.