Science and Technology news... :: View topic

jacko · Guest

well didn>t seem to post so again but with industrial one as starting
point (good deblock)

do above with auto level before to bitplanbe floyd dither

we get http://indi.hpsdr.com/image2.png

cheers do i get a prize? ;)

jacko · Guest

without the auto levels and just the other steps gives

http://indi.hpsdr.com/diffusiontop.png

ok

James Dow Allen · Guest

On Jun 10, 3:02 am, Thomas Richter <t...@math.tu-berlin.de> wrote:
[quote]AJ wrote:
I have an image which is very blocky (image can be seen at

Since you have unavoidably lost information, you need to reconstruct the
most probable input that generated your output (DCT coefficients, in
this term). This is the "maximum likelyhood" estimate.
[/quote]
I don>t know if Thomas implies this explicitly, but after
reducing blockiness (smoothing), it may be good to project the
result to the nearest point in the space of images consistent with
the compressed image. This is a very easy mathematical-like
step, though I don>t know if readily-available software does
this.

The steps (1) smooth; (2) project to nearest consistent image;
can be iterated, and even combined with other improvement
steps. (For example, a step to force pixels to a correct range,
e.g. 0 to 255 -- sometimes step (2) above will take pixels
out of range!)

James Dow Allen

Industrial One · Guest

On Jun 11, 8:22 pm, jacko <jackokr...@gmail.com> wrote:
[SNIP 3 consecutive posts]

lol... so smart, it took me a while to understand all of that text.
Jacko, not to be a dick, but how do you make intelligent posts yet
can>t spell? Is English your 2nd language, or you>re just chillaxed
and don>t care to be some faggot-ass geek that needs everything
perfect?

[quote]cheers do i get a prize? ;)
[/quote]
If you get your codec to work, send it to me and I>ll compress my 10
TB porn folder to a couple KB and promptly e-mail it to you. Deal?

jacko · Guest

On Jun 12, 12:12 pm, Industrial One <industrial_...@hotmail.com>
wrote:
[quote]On Jun 11, 8:22 pm, jacko <jackokr...@gmail.com> wrote:
[SNIP 3 consecutive posts]

lol... so smart, it took me a while to understand all of that text.
Jacko, not to be a dick, but how do you make intelligent posts yet
can>t spell? Is English your 2nd language, or you>re just chillaxed
and don>t care to be some faggot-ass geek that needs everything
perfect?
[/quote]
my spelling has often been quite bad. i had to resit my english o
level but was well upto and over a level computer science by 16 (not
that i>d sat an exam, cos you had to have and english o level to get
into university). i read a nice silver and black covered book on image
processing once, with all the fft and auto correlation etc. lots of
images in space and frequency equivelents. i think it was part of a
general series with other topics in other books. anyone got a copy cos
i think it was the best for the time. i was reading alot about APL at
the time, and also got to see the spectrum rom dissembly. i am very
laid back to the point that some people would say i didn>t care, but i
think i do care but have other impulses i care about too, so i tend
not to rush things beyond a true understanding of entia non sunt
multiplicanda preter necessitatum, i.e. keep it simple, but no simpler
than it NEEDS to be.

[quote]cheers do i get a prize? ;)

If you get your codec to work, send it to me and I>ll compress my 10
TB porn folder to a couple KB and promptly e-mail it to you. Deal?
[/quote]
10 TB of porn? umm I have to think about that ;) he he ha ha hoh ho ha
ha ha ha.

one final point is that i think an amiga may have extra filter modes
on the noise distribution front, that photo shop may not have. so this
slight noise addition reminds me of a hewlet packard audio compression
method that does low frequency mp3 and has a high frequency noise bust
matching. as the lower frequencies take less information.

a nice program to prequantize noise shape an image for jpg compression
may be useful.

cheers

AJ · Guest

On Jun 12, 2:53 pm, jacko <jackokr...@gmail.com> wrote:
[quote]On Jun 12, 12:12 pm, Industrial One <industrial_...@hotmail.com
wrote:

On Jun 11, 8:22 pm, jacko <jackokr...@gmail.com> wrote:
[SNIP 3 consecutive posts]

lol... so smart, it took me a while to understand all of that text.
Jacko, not to be a dick, but how do you make intelligent posts yet
can>t spell? Is English your 2nd language, or you>re just chillaxed
and don>t care to be some faggot-ass geek that needs everything
perfect?

my spelling has often been quite bad. i had to resit my english o
level but was well upto and over a level computer science by 16 (not
that i>d sat an exam, cos you had to have and english o level to get
into university). i read a nice silver and black covered book on image
processing once, with all the fft and auto correlation etc. lots of
images in space and frequency equivelents. i think it was part of a
general series with other topics in other books. anyone got a copy cos
i think it was the best for the time. i was reading alot about APL at
the time, and also got to see the spectrum rom dissembly. i am very
laid back to the point that some people would say i didn>t care, but i
think i do care but have other impulses i care about too, so i tend
not to rush things beyond a true understanding of entia non sunt
multiplicanda preter necessitatum, i.e. keep it simple, but no simpler
than it NEEDS to be.

cheers do i get a prize? ;)

If you get your codec to work, send it to me and I>ll compress my 10
TB porn folder to a couple KB and promptly e-mail it to you. Deal?

10 TB of porn? umm I have to think about that ;) he he ha ha hoh ho ha
ha ha ha.

one final point is that i think an amiga may have extra filter modes
on the noise distribution front, that photo shop may not have. so this
slight noise addition reminds me of a hewlet packard audio compression
method that does low frequency mp3 and has a high frequency noise bust
matching. as the lower frequencies take less information.

a nice program to prequantize noise shape an image for jpg compression
may be useful.

cheers
[/quote]

On Jun 11, 6:12 pm, aruzinsky <aruzin...@general-cathexis.com> wrote:
[quote]On Jun 9, 5:38 am, AJ <arandal...@gmail.com> wrote:

Hi All,

I have an image which is very blocky (image can be seen athttp://img222.imageshack.us/my.php?image=degradedimagesh2.png) and
would like to try and improve it>s quality as much as possible.

I understand that the image is of such a poor quality that I do not
expect any huge improvements - that is not my aim, my aim is to see if
it can be improved at all, and if so, which method(s) could yield the
best results.

So far I have tried some iterative restoration techniques such as
Richardson-Lucy deconvolution and Blind deconvolution - both have
proved to give a slight improvement.

I was hoping someone would have some other suggestions as to how I
could possibly improve the image.

Just in case understanding how the image was degraded in the first
place might help lead to a good suggestion for improvement algorithms,
I created the image using the following steps:

* 8x8 Sub-Block and DCT image
* store top 2 coefficients from each sub-block
* use these to create a new image

Many thanks,

AJ

Okay, this improved result,

http://www.general-cathexis.com/images/degradedimagesh2_4.jpg

, was obtained by incorporating some information from the cosine half
wave within each 8 pixel interval. The values are all of the same
sign within each half interval so I can get a fairly natural looking
image by reducing to 128 x 64 using a box kernel. Whereas, a 64 x 64
reduction averages the half cosine to zero, the 128 x 64 reduction
sort of makes it a half square wave.

Here is the reduced image in 16 bit format,

http://www.general-cathexis.com/images/degradedimagesh2_Reduced.png

, which I enlarged 8X using a proprietary nonlinear method and then
reduced the width 0.5X using box interpolation.
[/quote]
Just a quick thank you to everyone who has made the effort to
experiment with my degraded image, I really appreciate it! :)

I have measured the PSNR of each submitted (including some suggestions
from the 'sci.image.processing' group, which can be found here -
'http://groups.google.com/group/sci.image.processing/browse_thread/
thread/637db3e922109d18#'), against the original copy of the image.
The results are as follows:

Aruzinsky (http://www.general-cathexis.com/images/
degradedimagesh2_2.png) - 25.81 dB
Aruzinsky (http://www.general-cathexis.com/images/
degradedimagesh2_3.png) - 25.51 dB
Aruzinsky (http://www.general-cathexis.com/images/
degradedimagesh2_4.jpg) - 27.91 dB
pisz_na (http://i287.photobucket.com/albums/ll123/ememek/r23.png) -
19.27 dB
jacko (http://indi.hpsdr.com/degradedmagesh11.png) - 20.05 dB
jacko (http://indi.hpsdr.com/image2.png) - 19.37 dB
jacko (http://indi.hpsdr.com/diffusiontop.png) - 22.73 dB
mine (AJ) (http://i287.photobucket.com/albums/ll146/arandall85/
interpolated_iterated_image.png) 24.55 dB

You might get a prize jacko!
The only issue I have got is that ideally I would like to implement
the technique described in Matlab, since the rest of my image
processing techniques are written using it, and the process needs to
be automated, so unfortunately Photoshop is out of the question :(

Does anyone have any idea how I could go about converting jacko>s
method using Photoshop into a format I could perform using Matlab?:

Many thanks,

AJ

jacko · Guest

<snip>
[quote]Does anyone have any idea how I could go about converting jacko>s
method using Photoshop into a format I could perform using Matlab?:
[/quote]
Some idea but don>t know how exact this is

1) scale to twice size x and y, with the interpolated value depending
on the weighted sums of the four by four grid of pixels surrounding it
(i think this is bicubic).
2) quantize each pixel to 1 bit black or white and diffuse the
risidual error of the quantization into surrounding pixels at 1/12 for
a corner and 1/6 for a side. then see if any pixels changed their
final 1 bit quantization and repeat risidual diffusion. look up floyd
steinberg for the absolute best on this. This process diffuses the
block boundries which have been smoothed into each other but the image
looks VERY pixelated. but the noise is all above 4 pixels in
frequency, well most of it.
3) then blur it by bleeding pixels into each other, via a low pass
filter set to use only closest neighbours (i think thats what happens)
this is done twice as a third time has not much percievable
improvement (not any).
4) all the noise is in the high frequencies and so a weighted SNR
would benefit my methods.

hope this helped.

jacko · Guest

hi

took the winner and did http://indi.hpsdr.com/r23.png

jacko · Guest

http://indi.hpsdr.com/r23.png

jacko · Guest

On Jun 12, 6:56 pm, jacko <jackokr...@gmail.com> wrote:
> http://indi.hpsdr.com/degradedimagesh2_4.png

Industrial One · Guest

On Jun 12, 12:49 pm, jacko <jackokr...@gmail.com> wrote:
[quote]On Jun 12, 6:56 pm, jacko <jackokr...@gmail.com> wrote:

http://indi.hpsdr.com/degradedimagesh2_4.png
[/quote]
Dude, you posted 3x in a row, again.

jacko · Guest

Sorry to industrial one for making this a fourth post in a row, but I
should point out that if floyd-steinberg dithering is used to
preprocess an image to reduce the colour plane, then maybe a walsh
transform would suite the process better than a DCT.

just a thought.

cheers

jacko · Guest

And my final word on the matter is that I guess the phase of the walsh
functions at high frequencies (in the oversample range particularly)
do not matter as much as single pixel errors in location are less
objectionable than multi pixel mis location errors of the lower
frequencies. Also the huffmann overhead of variable length symbols is
avoided. A slight noise bleed is tolerable.

Maybe constraining dithering to horizontal scan lines helps with the
amount of oversampling needed, but maybe it doesn>t. Work with 1 bit
DACs suggests a 4 * oversample is sufficient.

cheers.

p.s. I say no more.

Industrial One · Guest

On Jun 12, 7:53 am, jacko <jackokr...@gmail.com> wrote:
[quote]On Jun 12, 12:12 pm, Industrial One <industrial_...@hotmail.com
wrote:

On Jun 11, 8:22 pm, jacko <jackokr...@gmail.com> wrote:
[SNIP 3 consecutive posts]

lol... so smart, it took me a while to understand all of that text.
Jacko, not to be a dick, but how do you make intelligent posts yet
can>t spell? Is English your 2nd language, or you>re just chillaxed
and don>t care to be some faggot-ass geek that needs everything
perfect?

my spelling has often been quite bad. i had to _resit_ my english o
level but was well upto and over a level computer science by 16 (not
that i>d sat an exam, cos you had to have _and_ english _o_ level to get
[/quote]
See what I mean? Are you seriously telling me that you can>t spell
"recite?" Did you truncate the other words on purpose? And its not
just the spelling, its the grammar -- discursive, misleading,
sometimes becomes overly eloquent outta nowhere etc. (although it>s
mostly on topic.)

Serious question now: do you have some slight autism?

[SNIP]
[quote]cheers do i get a prize? ;)

If you get your codec to work, send it to me and I>ll compress my 10
TB porn folder to a couple KB and promptly e-mail it to you. Deal?

10 TB of porn? umm I have to think about that ;) he he ha ha hoh ho ha
ha ha ha.
[/quote]
You stfu, it>s a real bargain. Think about it: no more scrounging
around printshops with their piss-poor dialup connections for the good
stuff, you get everything in a couple KB. Unless you>re not certain
that your algorithm can absorb 10 TB into 10 KB?

jacko · Guest

On Jun 14, 11:23 pm, Industrial One <industrial_...@hotmail.com>
wrote:
[quote]On Jun 12, 7:53 am, jacko <jackokr...@gmail.com> wrote:

On Jun 12, 12:12 pm, Industrial One <industrial_...@hotmail.com
wrote:

On Jun 11, 8:22 pm, jacko <jackokr...@gmail.com> wrote:
[SNIP 3 consecutive posts]

lol... so smart, it took me a while to understand all of that text.
Jacko, not to be a dick, but how do you make intelligent posts yet
can>t spell? Is English your 2nd language, or you>re just chillaxed
and don>t care to be some faggot-ass geek that needs everything
perfect?

my spelling has often been quite bad. i had to _resit_ my english o
level but was well upto and over a level computer science by 16 (not
that i>d sat an exam, cos you had to have _and_ english _o_ level to get

See what I mean? Are you seriously telling me that you can>t spell
"recite?" Did you truncate the other words on purpose? And its not
just the spelling, its the grammar -- discursive, misleading,
sometimes becomes overly eloquent outta nowhere etc. (although it>s
mostly on topic.)
[/quote]
re-sit: desk paper chair again -> to resit
o level: ordinary level paper usual age of sitting = 16

[quote]Serious question now: do you have some slight autism?
[/quote]
Never diagnosed, maybe I is a little rain man. The autistic society
used to be (maybe still is) just up the road from my old university

cheers


www.GetXFactor.com Leading Technology, Science, Agriculture News and information	Part of the Identityscape.com network... getxfactor.com jmoodmusic.com smartbusinesschoices.com mintdepot.com lowfaresalways.com evangelicalview.com shoppingpodder.com soproudlywehail.com webnews.ws currenthumor.com

Science and Technology news... Forum Index -> Compression Forum	Goto page Previous 1, 2, 3 Next
Page 2 of 3