[Insight-users] Linear Vs Non-linear filters
Neha D
itkneha at yahoo . com
Sun, 26 Oct 2003 15:50:18 -0800 (PST)
--0-142744199-1067212218=:77456
Content-Type: text/plain; charset=us-ascii
Hi Luis,
Thank you very much for your valuable information. I appreciate that.
Thanks,
Neha
Luis Ibanez <luis . ibanez at kitware . com> wrote:
Hi Neha,
Your list of question will generate good material
for a course in image processing.
The ITK theory book "Insight into Images" that will
be published soon will certainly give you a good
overview of these concepts.
In the meantime here are some comments:
1) Yes, you should consider if a filter is linear
or non-linear. In practice, linear filter are
only useful a preprocessing and postprocessing
filters. They lack the capabilities for performing
image analysis.
2) In a real application you will combine linear and
non-linear filters. Most people in image processing
is very pragmatic and will actually not care much
about whether the filter is linear or not, as long
as the filter contributes to produce the desired
result.
3) There are not frequency domain filters in ITK at
this point. This will certainly change with the
recent introduction of efficient FFT methods.
4) In practice, Segmentation is mostly a space-based
activity rather than a frequency one. You could
of course imagine cases in which pixel texture
could contribute to define anatomical regions,
and in such cases frequency-base analysis could
help to enhance the delineation of such regions.
The current trend in medical image analysis points
more in the direction of creating models that contain
anatomical information. In that trend, frequency based
filters have only a secondary role, as preprocessing
enhancement filters.
In other words, if you are planning what book to buy
next, an "Anatomy Atlas" should be top on your list
before any other image processing book.
Regards,
Luis
---------------
Neha D wrote:
> Hello All,
>
> While applying filters for specific purpose, should we consider if that
> filter is linear or non-linear. When to use linear and when to use
> non-linear filter?
>
> Also, among the filters in ITK, which are spatial filters and which one
> are frequency domain filters.
>
> What is more useful for medical images ? why ?
>
> thanks,
>
> neha
>
> ------------------------------------------------------------------------
> Do you Yahoo!?
> Yahoo! SiteBuilder
> - Free,
> easy-to-use web site design software
---------------------------------
Do you Yahoo!?
Yahoo! SiteBuilder - Free, easy-to-use web site design software
--0-142744199-1067212218=:77456
Content-Type: text/html; charset=us-ascii
<DIV>Hi Luis,</DIV>
<DIV> </DIV>
<DIV>Thank you very much for your valuable information. I appreciate that.</DIV>
<DIV> </DIV>
<DIV>Thanks,</DIV>
<DIV>Neha<BR><BR><B><I>Luis Ibanez <luis . ibanez at kitware . com></I></B> wrote:</DIV>
<BLOCKQUOTE class=replbq style="PADDING-LEFT: 5px; MARGIN-LEFT: 5px; BORDER-LEFT: #1010ff 2px solid"><BR>Hi Neha,<BR><BR>Your list of question will generate good material<BR>for a course in image processing.<BR><BR>The ITK theory book "Insight into Images" that will<BR>be published soon will certainly give you a good<BR>overview of these concepts.<BR><BR>In the meantime here are some comments:<BR><BR>1) Yes, you should consider if a filter is linear<BR>or non-linear. In practice, linear filter are<BR>only useful a preprocessing and postprocessing<BR>filters. They lack the capabilities for performing<BR>image analysis.<BR><BR>2) In a real application you will combine linear and<BR>non-linear filters. Most people in image processing<BR>is very pragmatic and will actually not care much<BR>about whether the filter is linear or not, as long<BR>as the filter contributes to produce the desired<BR>result.<BR><BR>3) There are not frequency domain filters in ITK at<BR>this point. This will
certainly change with the<BR>recent introduction of efficient FFT methods.<BR><BR>4) In practice, Segmentation is mostly a space-based<BR>activity rather than a frequency one. You could<BR>of course imagine cases in which pixel texture<BR>could contribute to define anatomical regions,<BR>and in such cases frequency-base analysis could<BR>help to enhance the delineation of such regions.<BR><BR>The current trend in medical image analysis points<BR>more in the direction of creating models that contain<BR>anatomical information. In that trend, frequency based<BR>filters have only a secondary role, as preprocessing<BR>enhancement filters.<BR><BR>In other words, if you are planning what book to buy<BR>next, an "Anatomy Atlas" should be top on your list<BR>before any other image processing book.<BR><BR><BR><BR>Regards,<BR><BR><BR><BR>Luis<BR><BR><BR><BR>---------------<BR>Neha D wrote:<BR>> Hello All,<BR>> <BR>> While applying filters for specific purpose, should we consider if
that <BR>> filter is linear or non-linear. When to use linear and when to use <BR>> non-linear filter?<BR>> <BR>> Also, among the filters in ITK, which are spatial filters and which one <BR>> are frequency domain filters.<BR>> <BR>> What is more useful for medical images ? why ?<BR>> <BR>> thanks,<BR>> <BR>> neha<BR>> <BR>> ------------------------------------------------------------------------<BR>> Do you Yahoo!?<BR>> Yahoo! SiteBuilder <BR>> <HTTP: evt="10469/*http://sitebuilder . yahoo . com" us.rd.yahoo.com>- Free, <BR>> easy-to-use web site design software<BR><BR><BR><BR></BLOCKQUOTE><p><hr SIZE=1>
Do you Yahoo!?<br>
<a href="http://us . rd . yahoo . com/evt=10469/*http://sitebuilder . yahoo . com">Yahoo! SiteBuilder</a> - Free, easy-to-use web site design software
--0-142744199-1067212218=:77456--