Science and Technology news... :: View topic

AJ · Guest

Hi All,

I have an image which is very blocky (image can be seen at
http://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

Industrial One · Guest

On Jun 9, 6:05 am, AJ <arandal...@gmail.com> wrote:
[quote]Hi All,

http://img222.imageshack.us/my.php?image=degradedimagesh2.png) and[/quote]

Hawt.

[quote]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.
[/quote]
Probably a bi-cubic interpolation engine, which blends pixels together
at the simplest. The more complex processes involve detecting curves,
linear shapes/luminescence patterns and restores some quality in
accordance with aforementioned characteristics.

[quote]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.
[/quote]
There>s only so much information you can "guess back" when you>ve
trashed a portion of it. If I ripped out every 5th letter of this
message, you>d probably be able to guess the missing characters 'cuz
you still have a majority of existing info, but if I took out every
2nd letter... some words you could restore, most you can>t cuz they
got multiple possibilities, your picture would be equal to something
like that.

[quote]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
[/quote]
DCT is beyond me. But your image was degraded by a lossy algorithm --
irreversible. Information was lost, plain and simple.

Just pay an experienced photoshop user to restore your picture by hand.

AJ · Guest

On Jun 9, 3:36 pm, Industrial One <industrial_...@hotmail.com> wrote:
[quote]On Jun 9, 6:05 am, AJ <arandal...@gmail.com> wrote:> Hi All,

http://img222.imageshack.us/my.php?image=degradedimagesh2.png) and

Hawt.

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.

Probably a bi-cubic interpolation engine, which blends pixels together
at the simplest. The more complex processes involve detecting curves,
linear shapes/luminescence patterns and restores some quality in
accordance with aforementioned characteristics.

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.

There>s only so much information you can "guess back" when you>ve
trashed a portion of it. If I ripped out every 5th letter of this
message, you>d probably be able to guess the missing characters 'cuz
you still have a majority of existing info, but if I took out every
2nd letter... some words you could restore, most you can>t cuz they
got multiple possibilities, your picture would be equal to something
like that.

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

DCT is beyond me. But your image was degraded by a lossy algorithm --
irreversible. Information was lost, plain and simple.

Just pay an experienced photoshop user to restore your picture by hand.
[/quote]
Thanks for your reply.

Paying an experienced photshop user is not really an option...I have a
copy of the original image, I would just use that if that was my aim.
What I am trying to do is restore this poor image as much as possible
before finally using it as a form of highly compressed restoration
data if portions of the original image are degraded.

Good suggestion with the interpoliation. I have just experimented
with that and have got some reasonable results.
After some iterative restoration and interpolation, the image is
looking quite a lot better :) Do you have any suggested algorithms/
techniques I could try and apply to fix the edges (which seem to be
the worst regions of the image). Possibly some kind of edge
detection, followed by a filter to blur only the selected regions? I
am currently experimenting in Matlab if that makes any difference
(maybe there are already some implemented functions to complete this
task? I have had a google but couldn>t find anything...).

Many thanks,

AJ

Sportman · Guest

AJ wrote:
[quote]Paying an experienced photshop user is not really an option...I have a
copy of the original image, I would just use that if that was my aim.
What I am trying to do is restore this poor image as much as possible
before finally using it as a form of highly compressed restoration
data if portions of the original image are degraded.
You can try http://vectormagic.com a web browser tool what can make[/quote]
little blocks more smooth.

AJ · Guest

On Jun 9, 5:19 pm, Sportman <sport...@gmail.com> wrote:
[quote]AJ wrote:
Paying an experienced photshop user is not really an option...I have a
copy of the original image, I would just use that if that was my aim.
What I am trying to do is restore this poor image as much as possible
before finally using it as a form of highly compressed restoration
data if portions of the original image are degraded.

You can tryhttp://vectormagic.coma web browser tool what can make
little blocks more smooth.
[/quote]
Thanks for your response. I gave it a go but unfortunately it didn>t
give much in the way of improvements.

AJ · Guest

On Jun 9, 9:02 pm, Thomas Richter <t...@math.tu-berlin.de> wrote:
[quote]AJ wrote:
Hi All,

I have an image which is very blocky (image can be seen at
http://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

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. One good
criterion for this should be that the bottom line of one 8x8 block
should have the same (or very close) value to the top-line of the block
below. That is, for all possible combinations of missing DCT
coefficients (namely those you got rid of), find the combination that
minimizes the error between the last line of this block and the top line
of the block below.

I know that this doesn>t yet formulate a reasonable algorithm because it
is very slow, but it can surely be turned into something more workable
with a bit of mathematics - and it is clearly the "optimal" solution for
this problem given the assumed "smoothness" of the input picture.

Other than that, you>re of course free to pick any heuristic that seems
reasonable for you, i.e. detect edges by checking for the gradient and
smooth out in vertical direction otherwise.

So long,
Thomas
[/quote]
That sounds like a good idea, and I will look into it. In the
meantime, here is the best restoration attempt I have got so far...

http://i287.photobucket.com/albums/ll146/arandall85/interpolated_iterated_image.png

To get to this point, I have used blind deconvolution, and then
interpolated the result (thanks 'Industrial One'). I>m quite happy
with this outcome, although it is a little blurry. What do you all
think?

Ideally, I would like to "fix" the edges in some way, as they are
still a bit blocky, and I would like to sharpen the whole image
slightly - but I think I may be hoping for a bit too much! ;)

If anyone has any further suggestions to improve this image, I would
be really grateful. Nearly there...

Kind regards,

AJ

Thomas Richter · Guest

AJ wrote:
[quote]Hi All,

I have an image which is very blocky (image can be seen at
http://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
[/quote]
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. One good
criterion for this should be that the bottom line of one 8x8 block
should have the same (or very close) value to the top-line of the block
below. That is, for all possible combinations of missing DCT
coefficients (namely those you got rid of), find the combination that
minimizes the error between the last line of this block and the top line
of the block below.

I know that this doesn>t yet formulate a reasonable algorithm because it
is very slow, but it can surely be turned into something more workable
with a bit of mathematics - and it is clearly the "optimal" solution for
this problem given the assumed "smoothness" of the input picture.

Other than that, you>re of course free to pick any heuristic that seems
reasonable for you, i.e. detect edges by checking for the gradient and
smooth out in vertical direction otherwise.

So long,
Thomas

Industrial One · Guest

On Jun 9, 3:38 pm, AJ <arandal...@gmail.com> wrote:
[quote]That sounds like a good idea, and I will look into it. In the
meantime, here is the best restoration attempt I have got so far...

http://i287.photobucket.com/albums/ll146/arandall85/interpolated_iter...

To get to this point, I have used blind deconvolution, and then
interpolated the result (thanks 'Industrial One'). I>m quite happy
with this outcome, although it is a little blurry. What do you all
think?
[/quote]
I think its hawt despite the degredation. Send us nudes xD

[quote]Ideally, I would like to "fix" the edges in some way, as they are
still a bit blocky, and I would like to sharpen the whole image
slightly - but I think I may be hoping for a bit too much! ;)
[/quote]
You can selectively smoothen out regions of a picture with any
photoshop app (Adobe, Corel) given some patience and work. But again,
this is a task beyond a restoration app and falls into the category of
photoshopping by hand, which is the only plausible way to restore the
image completely.

AJ · Guest

On Jun 10, 6:30 am, Industrial One <industrial_...@hotmail.com> wrote:
[quote]On Jun 9, 3:38 pm, AJ <arandal...@gmail.com> wrote:

That sounds like a good idea, and I will look into it. In the
meantime, here is the best restoration attempt I have got so far...

http://i287.photobucket.com/albums/ll146/arandall85/interpolated_iter...

To get to this point, I have used blind deconvolution, and then
interpolated the result (thanks 'Industrial One'). I>m quite happy
with this outcome, although it is a little blurry. What do you all
think?

I think its hawt despite the degredation. Send us nudes xD

Ideally, I would like to "fix" the edges in some way, as they are
still a bit blocky, and I would like to sharpen the whole image
slightly - but I think I may be hoping for a bit too much! ;)

You can selectively smoothen out regions of a picture with any
photoshop app (Adobe, Corel) given some patience and work. But again,
this is a task beyond a restoration app and falls into the category of
photoshopping by hand, which is the only plausible way to restore the
image completely.
[/quote]
Nudes as requested...

http://www.lenna.org/full/len_hires.html

(Bet you weren>t expecting that request to really materialise huh!).

I guess I will just wait and see if anyone else has any other
suggestions.

Kind regards and thank you to everyone who helped.

George Johnson · Guest

"AJ" <arandall85@gmail.com> wrote in message
news:8d7af91c-9dc8-490b-8cd6-25fe724be9a6@27g2000hsf.googlegroups.com...
[quote]Hi All,

I have an image which is very blocky (image can be seen at
http://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
[/quote]
I can do you one better. I can give you Lena (the name of the image) in
color and multiple resolutions.
http://www.codinghorror.com/blog/archives/000794.html

http://www.codinghorror.com/blog/images/lena512color.jpg

In fact, here is the "rest of the story".
http://www.cs.cmu.edu/~chuck/lennapg/
Introduction
The Lenna (or Lena) picture is one of the most widely used standard
test images used for compression algorithms. The comp.compression FAQ says
the following:

For the curious: 'lena' or 'lenna' is a digitized Playboy
centerfold, from November 1972. (Lenna is the spelling in Playboy, Lena is
the Swedish spelling of the name.) Lena Soderberg (ne Sjööblom) was last
reported living in her native Sweden, happily married with three kids and a
job with the state liquor monopoly. In 1988, she was interviewed by some
Swedish computer related publication, and she was pleasantly amused by what
had happened to her picture. That was the first she knew of the use of that
picture in the computer business.
For the full details on the history of this image check out this
excellent May 2001 article in the Newsletter of the IEEE Professional
Communication Society by Jamie Hutchinson. Here>s an excerpt:

Alexander Sawchuk estimates that it was in June or July of 1973 when
he, then an assistant professor of electrical engineering at the USC Signal
and Image Processing Institute (SIPI), along with a graduate student and the
SIPI lab manager, was hurriedly searching the lab for a good image to scan
for a colleague>s conference paper. They had tired of their stock of usual
test images, dull stuff dating back to television standards work in the
early 1960s. They wanted something glossy to ensure good output dynamic
range, and they wanted a human face. Just then, somebody happened to walk in
with a recent issue of Playboy.
The engineers tore away the top third of the centerfold so they
could wrap it around the drum of their Muirhead wirephoto scanner, which
they had outfitted with analog-to-digital converters (one each for the red,
green, and blue channels) and a Hewlett Packard 2100 minicomputer. The
Muirhead had a fixed resolution of 100 lines per inch and the engineers
wanted a 512 x 512 image, so they limited the scan to the top 5.12 inches of
the picture, effectively cropping it at the subject>s shoulders.

The original image is still available as part of the USC SIPI Image
Database in their "miscellaneous" collection.

Over the years there has been quite a bit of controversy over the use
of this image. Some people proposed banning the use of this image because of
its source. Also, Playboy threatened to prosecute the unauthorized use of
the image. Check out an editorial by the editor of SPIE journal Optical
Engineering. Check out a note by the former editor-in-chief of the IEEE
Transactions on Image Processing. According to Wired Magazine, Playboy has
eased up in its pursuit of the copyright violators of this image.

Another interesting piece of trivia is that Lenna>s issue (November
1972) was Playboy>s best selling issue ever and sold 7,161,561 copies.

Don>t forget to check out Lenna>s birthday page (3/31) and playmate
directory page at Playboy. (Warning: contains nudity.)

There>s even a sonnet dedicated to Lenna also found here.

Industrial One · Guest

On Jun 10, 4:01 am, AJ <arandal...@gmail.com> wrote:
[quote]Nudes as requested...

http://www.lenna.org/full/len_hires.html

(Bet you weren>t expecting that request to really materialise huh!).

I guess I will just wait and see if anyone else has any other
suggestions.

Kind regards and thank you to everyone who helped.
[/quote]
Well played.

jacko · Guest

On Jun 9, 1:05 pm, AJ <arandal...@gmail.com> wrote:
[quote]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
[/quote]
as an initial attempt, i>d oversample the image to 4*4 it size, and
run it bicubic to get the best curvature/alias blend. then i>d do a
deconvolution/blur removal, cos in some senses the low pass effect by
dropping the dct coefficients as low frequencies would be the largest
on average. the a slight high pass/edge detect or .FIF with 30 mins of
user feedback on genetic algorithm selection of a good transform set.
(well maybe not) he.he.

no i would have done the 4*4 biqubic and a 3 bit colour plane floyd
steinberg before anything else to get some compression. although this
may have added to the file size (unlike 1d audio noise shaping) it
would move most of the noise into the high frequency for just a small
increment in file size. if the floyd stienberg was slightly modified
to adjust for dct correlation?? who knows. a one coefficient dct at
this 4*4 resolution (and lower colour planes) may give a good one??

na really a bicubic grow and a sharpen filter is the best i can do ?

jacko · Guest

make that a 1 bit plane colour as tis black and white, me must be
blind sometimes ;)

so maybe a 2*2 floyd colour reduction gives a easy one coefficient
dct??

jacko · Guest

i did this in photoshop

image size 2*2 grow bicubic
auto levels
make bitmap using defuse (floyd-steinberg)
back to grey scale
blur
blur
reduce 2*2 back to origional size

looks better as somehow the noise adds texture. and the double canvas
size somehow stops some of the pixelation

http://indi.hpsdr.com/degradedmagesh11.png

ok

jacko · Guest

but using industrial one>s deblur and doing the same except switch
auto levels before to bitmap you get

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


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 1, 2, 3 Next
Page 1 of 3