Re: Scientifically Based Presharpening for Enlargement



aruzinsky wrote:

So show us an example of an image that is either
*direct from a sensor, with no processing applied*, or at least a
simulation of such an image. (For example, you could take a much higher
resolution image and convolve/downsample it to get your example).
-------------------------------------------------
Why do you assume it is not such a simulation? In fact it is.


So, if I understand, your demonstration involves taking a high
resolution image and simulating a sensor array by using a box-car like
filter to downsample. After that you investigate methods of recovering
the original image.

The general premise is correct. The box-car like convolution of the
sensor array has a known frequency response and a sharpening filter when
applied goes along way to reverse (it has the inverse frequency
response) the effects of the sampling. So in theory, if you start with a
appropriately bandlimited image and you carry out this exercise you
should be able to come very close to recovering the original image.
The problem is that in real life, and in your example too, the original
is not "appropriately bandlimited" and thus when the simulated sampling
is done aliasing occurs. That aliasing is permanent and irreversible.

What that means is, in order for your experiment to work, you need to
start with a high resolution blurry image (appropriately bandlimited so
that no aliasing occurs when you resample it) and then (after simulation
of capturing with a sensor array) your pre-sharpening technique will do
a better job of recovering the original blurry image than interpolation
without it.

-jim








-------------------------------------------------
What you have on the page at the moment shows really horrendous sharpening
artifacts (halos) as well as compression artifacts that completely
overwhelm the small difference between the two cases that you are trying
to show. Your starting images should be uncompressed and unsharpened.
-------------------------------------------------
There are no compression artifacts in the input images because PNG uses
lossless compression. The differences are obvious to my eyes. If you
don't see the differences you don't need presharpening.

-------------------------------------------------
Also, if the "data dependent Lanczos" filter is what's responsible for the
weird wave-like textures in the road surface at the bottom of the
image, don't use this filter! Adding artifacts that weren't there in the
original is generally a bad idea, and especially bad in something
that's supposed to be an example. Why not use bicubic polynomial
interpolation, which (a) is familiar to many people, and (b) doesn't
create such artifacts. Yes, it doesn't preserve high frequencies as well
as Lanczos, but you should still be able to see the difference
between the images with and without presharpening applied, and that's the
point of the example.
-------------------------------------------------
a. I provided the input images so people could test them with their
favorite enlargement method.

b. That "weird wave-like texture" is less offensive to the eyes than the
jagged edges you will get with bicubic and the halos will be similar with
both. Providing an exotic example without jagged edges illustrates the
versatility of the method.

-------------------------------------------------
3. You might address why the sharpening should be a preprocessing step. If
the sharpening is a linear filter, and interpolation is a linear
filter (and most popular interpolation methods are linear) then it should
not matter what order you perform them in - as long as you deal
with the change in magnification for the sharpening step. You might even
find a way to merge the sharpening into the interpolation.
On the other hand, if the interpolation filter is actually nonlinear, then
that's a reason why the sharpening needs to be a preprocess.
Dave
-------------------------------------------------
Although convolution is linear and commutative, linear operations are not
generally commutative. And, my Data Dependent Lanczos enlargement method
is not linear. An advantage of presharpening too obvious to mention is
that a smaller image can be sharpened more quickly than after enlargement.


----== Posted via Newsfeeds.Com - Unlimited-Unrestricted-Secure Usenet News==----
http://www.newsfeeds.com The #1 Newsgroup Service in the World! 120,000+ Newsgroups
----= East and West-Coast Server Farms - Total Privacy via Encryption =----
.

Relevant Pages

  • Re: Scientifically Based Presharpening for Enlargement
    ... sharpening artifacts as well as compression artifacts that ... image, don't use this filter! ... You might address why the sharpening should be a preprocessing step. ... If the sharpening is a linear filter, and interpolation is a linear ...
    (sci.image.processing)
  • Re: unsharp mask?
    ... And I am disagreeing with your statements that it always results in sharpening artefacts and a mess - that certainly is NOT true. ... I have yet to see a *single* instance of "sharpening artefacts" caused by the application of the standard level of the sharpening filter even to images downsampled from raw scans. ...
    (comp.periphs.scanners)
  • Re: active PFC input filter
    ... I have a simulation of multiphase active power factor correction ... There are three phases with each phase having a 75uH/25Amp inductor. ... The input filter is just an LC lowpass on the output of the bridge ...
    (sci.electronics.design)
  • Baseband Pulse Shaping QPSK in IQ
    ... for this basic pulse shaping problem. ... I am doing a simulation of a QPSK system and I understand the basic ... concept of pulse shaping and matching filtering. ... Q path with the same filter coefficients? ...
    (comp.dsp)
  • Body CS Problem
    ... I am simulating a system with a few rigid bodies. ... I connected a body sensor to one of the bodies and selected "with ... respect to the coordinate system" selection as "Local. ... "Absolute" selected and run the simulation again. ...
    (comp.soft-sys.matlab)