home *** CD-ROM | disk | FTP | other *** search
- Xref: sparky comp.graphics:11654 comp.multimedia:3303 comp.compression:3782
- Path: sparky!uunet!ferkel.ucsb.edu!taco!gatech!swrinde!sdd.hp.com!caen!uwm.edu!biosci!agate!stanford.edu!eos!jbm
- From: jbm@eos.arc.nasa.gov (Jeffrey Mulligan)
- Newsgroups: comp.graphics,comp.multimedia,comp.compression
- Subject: Re: Looking for analytic image quality metric
- Message-ID: <1992Nov9.221547.29540@eos.arc.nasa.gov>
- Date: 9 Nov 92 22:15:47 GMT
- References: <1dequfINNlj4@sixgun.East.Sun.COM>
- Distribution: usa
- Organization: NASA Ames Research Center
- Lines: 44
-
- westover@sunpix.East.Sun.COM (Lee Westover - Sun NC Development Center) writes:
-
-
- >Hello netters,
-
- >We are doing some work with digital video compression/decompression and
- >are looking for a way to measure image quality that is not subjective.
- >We would like to find a reasonable way of comparing an original image
- >with the decompressed compressed image. MSE and MAE are bad metrics
- >and we do not want to hassle with user studies.
-
- >Does anybody know of a good image quality measure?
-
- Different types of measures are needed depending on what question you want
- to ask: ie., a different computation is required to answer "are the
- artifacts visible or invisible" as opposed to "of these two images,
- both with visible artifacts, which is better."
-
- A simple model to predict the visibility of artifacts is to take
- the error image, filter it with a filter that mimics human contrast
- sensitivity, and then either integrate over the image or look for
- the max. More sophisticated models incorporate masking effects,
- oriented subbands, and other stuff.
-
- > Know of any
- >institutions working on finding a good analytic image quality measure?
- >Know of any papers about this subject?
-
- People that I know are working in this area:
-
- Peter Baarten (consultant, Netherlands)
- Jeff Lubin (Sarnoff Research Lab)
- Scott Daly (Kodak Research)
-
- Places to look for papers are proceedings of the Society for Information
- Display (SID), and recent SPIE conferences on "Human vision and electronic
- imaging."
-
-
- --
-
- Jeff Mulligan (jbm@vision.arc.nasa.gov)
- NASA/Ames Research Ctr., MS 262-2, Moffett Field CA, 94035-1000
- (415) 604-3745
-