Re: Error Diffusion Artifacts
sloan@cis.uab.edu (Kenneth R. Sloan)
8 Jun 1997 01:46:14 GMT
In article <5nb8dj$21l@sunsystem5.informatik.tu-muenchen.de>,
Andi Braun wrote:
>How can I solve the problem of artifacts in error diffusion
>dithering? Does rescaling the error (Floyd) work? How can I
>do this. What about using 12 error fractions?
The most effective means that I have encountered involves combining
error-diffusion with ordered dither. The idea is to pick an ordered
dither pattern which is "nice", and use it at very low amplitude. This
perturbs the patterns which can arise in error diffusion (the primary
effect - which you can get by adding any sort of noise) and also
provides a "nice" pattern to emerge in regions where *some* sort fo
pattern *must* emerge (large, nearly-constant regions).
The key idea is to add the ordered dither (look at it as a very
structured "pseudo-random" noise) before thresholding (choosing a
display color) BUT to use the original (non-noised) image values when
doing error diffusion. At each output pixel:
output(x,y) = Threshold(Image(x,y)+Pattern(u,v)+ErrorIn(x,y))
ErrorOut(x,y) = Image(x,y)-output(x,y)
...distribute ErrorOut as usual to other pixels' ErrorIn
Design handles are the Pattern, and the amplitude of the Pattern. Crank
up the Amplitude and you get pure Ordered Dither. Turn is way down, and
you get pure error-diffusion. In the middle, you get...
I have a couple of SPIE abstracts on this, many years ago. You might
search for them - but the above summary has most of the content.
"when texture is inevitable, impose a pleasing texture"
--
Kenneth Sloan sloan@cis.uab.edu
Computer and Information Sciences (205) 934-2213
University of Alabama at Birmingham FAX (205) 934-5473
Birmingham, AL 35294-1170 http://www.cis.uab.edu/info/faculty/sloan/
BarGold Electronics - PC & DSP Image Processing
bgold@netvision.net.il (Moshe Goldstein)
Sun, 08 Jun 1997 10:54:03 GMT
Dear Sir,
Thank you for your interest in BarGold's products.
Recently you have contact us by Email requesting for more information
about BarGold and its image processing products.
In order to be able to mail you our products description, pricing,
distributors etc. we need your fax and address.
Our WEB site URL is: http://www.bargold.com
Following please find some general information about BarGold.
Best regards,
Moshe Goldstein.
---------------------------------------------------------------------
BarGold Electronics Ltd. Develops, manufactures and markets Machine
Vision DSP based Real-time image Processing Systems. 1-5 DSPs 1-3
frame grabbers on board.
BarGold product lines are powerful compact vision systems, single-slot
PC boards, combining dedicated hardware and software, to obtain
unrivaled speed and versatility.
The systems operate inside a PC or without a PC in Stand-Alone mode.
Application software development is simplified by using a supplied
comprehensive image processing library (VIPS) including more than 500
subroutines.
Applications include complex vision tasks such as:
==================================================
- Inspection, non-contact measurements, quality control
- Assembly, Robot Guidance, Pick and Place machines,
- Sorting and Identification systems.
- PCB board testing, components location and verification,
- Surveillance, Security control systems,
- Video tracking for industrial and military applications.
- 50-250 MIPS using 1-5 DSPs on board.
---------------------------------------------------------------
COMPANY PROFILE
VISION FOR INDUSTRY, MEDICAL AND SCIENTIFIC PURPOSES.
ABOUT THE BUSINESS:
Real-Time Image Processing technology.
BarGold produces its own hardware and software for OEM products,
turnkey projects, or End-Users.
BarGold focuses on add-on H/W for PCs, or for Stand-Alone uses, where
both H/W and S/W are carefully tuned for optimal image processing &
Machine Vision, achieving: power, speed and unbeatable
cost-performance.
BarGold provides adaptable, friendly and versatile tools for actuating
robot and robot and machine devices. BarGold's hi-tech developments
are, leading-edge, state-of-the-art and aggressive.
BarGold's product lines are extremely powerful, very compact vision
systems, single-slot PC boards, combining dedicated hardware and
programmable software, to obtain unrivaled speed and versatility. The
systems operate in an IBM PC (or compatible) or in a stand-alone mode
after completion of application S/W writing and debugging (using
BarGold script or "C" language).
ABOUT THE ORGANIZATION:
Founded in 1988 by David Barkan and Moshe Goldstein, partners and
major shareholders (77.5%).
I.R.D.C. Ltd. (Israel Research & Development Corporation) (12.5%). (A
company owned by the Industrial Development Bank Limited of Israel).
Glenco Trading Ltd. Ireland (10%).
BOARD OF DIRECTORS
Mr. David Barkan, President.
Mr. Moshe Goldstein, General Manager and head of R&D.;
Mr. David Blass Director.
Mr. Hezy Ram, Director and CEO.
Situated north of Haifa, the company consists of highly skilled R & D
engineers, as well as a motivated sales and marketing team.
Success on the local market has been followed by expanding our export
markets either by joint ventures or by appointing local distributors
in Europe, North America & Canada, the Far East, Australia, South
America, South Africa.
-------------------------------------------
BarGold Electronics LTD.
7 Shlomo Ben Yoseph Street,
P.O.Box 25045, Haifa 31250, ISRAEL
TEL: 972-4-8414214
FAX: 972-4-8721079
Email: bargold@netvision.net.il
URL : http://www.bargold.com
------------------------------------------
Estimating woody biomass with Landsat MSS
"Jeroen Ticheler"
Sun, 8 Jun 1997 22:44:55 +0200
Hello,
I'm looking for researches that where conducted on the estimation of woody
biomass trough analysis of Landsat MSS images, especially for drier zones
(savannahs).
I'm working on a multitemporal analysis on changes in the amount of woody
plants in some parts of West-Africa, I would like to use a proofed method
to estimate this amount.
Furthermore I am interested in possible sources for additional Landsat
MSS
images (Bands 5, 6 and 7!?), especially the older ones (between 1972 and
1980), covering Eastern Burkina Faso (North and East of Ouagadougou) and
South-Eastern Mali (Dogon area)
Every suggestion is welcome and appreciated,
Thanks,Jeroen
__
Jeroen Ticheler
Student Tropical Forestry and GIS, Wageningen Agricultural University
6701 DM 14, Netherlands
JTicheler@Compuserve.com
Indepth review of Questar's World Construction Set 2.0
r_moncrf@ix.netcom.com(Roger Moncrief )
8 Jun 1997 20:42:11 GMT
I apologize for the cross posting however I believe this information
may be of interest to some of the readers of this newsgroup.
An indepth review of Questar's World Construction Set 2.0 is now on my
page. This terrain building, animation software works with both 3D
Studio Max and Lightwave. It can also use terrains created in Bryce's
terrain editor. I describe these procedures in the last section of the
review. It can animate sun rises, sunsets, moon rises, moon sets,
tides, waves, clouds, etc. It is currently available for Windows 95,
Windows NT, Amiga and DEC Alpha. Versions for the MacIntosh and
Unix/SGI are promised.
Roger A. Moncrief
http://www.indrev.com
Indepth Reviews
Indepth reviews of Bryce 2, Ray Dream Studio, Detailer, etc. on line
now.
Re: x86 processed grabber cards
Maxim Bohlmann
Sun, 08 Jun 1997 14:04:10 -0700
Hello Sheldon,
In your interesting note you made an intriguing statement : "The
TMS320C60 DSP makes a mighty
fine codec processor and the like for telecommuications; but, its not
really suited for imaging."
Could you please clarify why the C60 is NOT really suited for imaging.
Thanks,
Maxim Bohlmann
Sheldon L. Epstein wrote:
> I, for one, would be happy with a souped-up version of the
> TMS320C40 DSP; but, it doesn't appear that TI is going
> in that direction. The TMS320C60 DSP makes a mighty
> fine codec processor and the like for telecommuications;
> but, its not really suited for imaging. For that reason,
> I'm encouraging our vendor to look at MMX technology for
> the next edition.
>
> Thanks for considering my 2 cents.
>
> Shel@k9ape.com
Indepth review of Questar's World Construction Set
r_moncrf@ix.netcom.com(Roger Moncrief )
8 Jun 1997 21:49:34 GMT
An indepth review of Questar's World Construction Set 2.0 is now on my
page. This terrain building, animation software works with both 3D
Studio Max and Lightwave. It can also use terrains created in Bryce's
terrain editor. I describe these procedures in the last section of the
review. It can animate sun rises, sunsets, moon rises, moon sets,
tides, waves, clouds, etc. It is currently available for Windows 95,
Windows NT, Amiga and DEC Alpha. Versions for the MacIntosh and
Unix/SGI are promised.
Roger A. Moncrief
http://www.indrev.com
Indepth Reviews
Indepth reviews of Bryce 2, Ray Dream Studio, Detailer, etc. on line
now.
Indepth Review of Questar's World Construction Set
Roger Moncrief
Sun, 08 Jun 1997 20:19:02 GMT
I apologize for the cross posting however I believe this information may be of interest to some of the readers of this newsgroup.
An indepth review of Questar's World Construction Set 2.0 is now on my page. This terrain building, animation software works
with both 3D Studio Max and Lightwave. It can also use terrains created in Bryce's terrain editor. I describe these procedures in
the last section of the review. It can animate sun rises, sunsets, moon rises, moon sets, tides, waves, clouds, etc. It is currently
available for Windows 95, Windows NT, Amiga and DEC Alpha. Versions for the MacIntosh and Unix/SGI are promised.
Roger A. Moncrief
http://www.indrev.com
Indepth Reviews
Indepth reviews of Bryce 2, Ray Dream Studio, Detailer, etc. on line now.
Re: x86 processed grabber cards
"Sheldon L. Epstein"
Sun, 08 Jun 1997 19:26:40 -0500
Maxim Bohlmann wrote:
>
Hello Max & Friends,
My time schedule doesn't permit me to go into too much detail about
what I learned ftom TI's C60 Show-&-Tell; but, the following points
are important:
1. Not assembly program friendly - TI almost insists
on using their C Compiler development software because
dual accumulator architecture requires synchronizing
instructions. Our experience with C Compilers is that
the run 4 - 10 times slower than the native code we
write. That means 1.6 BOPS now translates to 160 MOPS.
2. DSPs like the C40 have auxiliary registers and
pointers make imaging a pleasure. For example, C40
pointers can use indirect addressing so that going
up and down or sideways in an image is easy. They
also perform modulo arithmatic so that you can
build kernels, ring-buffers and the like of any
size (not just 2^n) and reprogram them on the fly.
The C60 requires multiple instructions to do the
same thing. Probably a speed penalty of 4 in
assembler - maybe more in C.
Taken all together, I don't forsee the C60 offering much of
a performance improvement over the C40 for imaging.
Now when you move the C60 out of the two-dimensional world
to one-dimensional time-series problems in telecommunications,
then they blow DSPs like the C40 out of the water. But,
that is U.S. Robotics et al business - not ours. TI did
the right thing for its shareholders; but, it won't help
us. I seriously doubt if any of the major imaging card
builders will introduce a C60-based board.
Best Regards,
Shel@k9ape.com
> Hello Sheldon,
>
> In your interesting note you made an intriguing statement : "The
> TMS320C60 DSP makes a mighty
> fine codec processor and the like for telecommuications; but, its not
> really suited for imaging."
>
> Could you please clarify why the C60 is NOT really suited for imaging.
>
> Thanks,
>
> Maxim Bohlmann
>
> Sheldon L. Epstein wrote:
>
> > I, for one, would be happy with a souped-up version of the
> > TMS320C40 DSP; but, it doesn't appear that TI is going
> > in that direction. The TMS320C60 DSP makes a mighty
> > fine codec processor and the like for telecommuications;
> > but, its not really suited for imaging. For that reason,
> > I'm encouraging our vendor to look at MMX technology for
> > the next edition.
> >
> > Thanks for considering my 2 cents.
> >
> > Shel@k9ape.com
Estimating woody biomass with Landsat MSS
"Jeroen Ticheler"
Sun, 8 Jun 1997 22:44:55 +0200
Hello,
I'm looking for researches that where conducted on the estimation of woody
biomass trough analysis of Landsat MSS images, especially for drier zones
(savannahs).
I'm working on a multitemporal analysis on changes in the amount of woody
plants in some parts of West-Africa, I would like to use a proofed method
to estimate this amount.
Furthermore I am interested in possible sources for additional Landsat
MSS
images (Bands 5, 6 and 7!?), especially the older ones (between 1972 and
1980), covering Eastern Burkina Faso (North and East of Ouagadougou) and
South-Eastern Mali (Dogon area)
Every suggestion is welcome and appreciated,
Thanks,Jeroen
__
Jeroen Ticheler
Student Tropical Forestry and GIS, Wageningen Agricultural University
6701 DM 14, Netherlands
JTicheler@Compuserve.com
