| View previous topic :: View next topic |
| Author |
Message |
Moschops
Guest
|
 Posted: Thu Oct 02, 2008 8:41 pm Post subject: Quantitative measure of degreee of focus across an image, fo |
 |
|
I>m looking at various ways of segmenting images; one such idea I>ve had by
looking at the images in question is to identify areas of the image that are
in focus.
As such, I have thought of two ways of calculating some numbers.
Firstly, apply a sharpening filter to the whole image and then splitting the
image into regions, say 10x10 pixels, and assigning each a value based on
the difference between the original image and the sharpened image - the
larger the number, the greater the difference and thus the less in focus the
original image was in that region. I>d have to assume some things about the
nature of the focus, and would probably start with the assumption of a
gaussian PSF.
Secondly, applying edge detection across the whole image; I suppose that the
object in focus will have a sharp edge against the background, so if I
assign a value to every edge detected based on the gradient of the edge
detected, and then again break the image into regions and assign a value per
region based on the gradients of the edges detected (perhaps an average
gradient of edge detected) I would expect to see high values at the edges of
the object in focus.
By these two methods I hope to identify the region of an image that is in
focus, and direct my segmentation efforts accordingly.
I>d be very interested to hear the thoughts on these from people with
experience. I anticipate that there>s nothing new I can think of that hasn>t
been done already, so if anyone can direct me towards the results from
similar approaches I>d be very grateful.
'Chops |
|
| |
|
 Back to top |
bugbear
Guest
|
| Posted: Thu Oct 02, 2008 8:53 pm Post subject: Re: Quantitative measure of degreee of focus across an image |
|
|
Moschops wrote:
[quote]I>m looking at various ways of segmenting images; one such idea I>ve had by
looking at the images in question is to identify areas of the image that are
in focus.
[/quote]
People doing "focus stacking" (must?) have looked into this.
http://en.wikipedia.org/wiki/Focus_stacking
BugBear |
|
| |
|
| Back to top |
Moschops
Guest
|
| Posted: Thu Oct 02, 2008 8:56 pm Post subject: Re: Quantitative measure of degreee of focus across an image |
|
|
----- Original Message -----
From: "bugbear" <bugbear@trim_papermule.co.uk_trim>
Newsgroups: sci.image.processing
Sent: Thursday, October 02, 2008 4:53 PM
Subject: Re: Quantitative measure of degreee of focus across an image, for
segmentation. Your thoughts?
[quote]Moschops wrote:
I>m looking at various ways of segmenting images; one such idea I>ve had
by looking at the images in question is to identify areas of the image
that are in focus.
People doing "focus stacking" (must?) have looked into this.
http://en.wikipedia.org/wiki/Focus_stacking
BugBear
[/quote]
Could well have - I>m so new to this that I just don>t know these terms and
frequently am frustrated, knowing that people simply must have done these
things, but not knowing what they>re called! Focus stacking - I>ll look into
it. Many thanks.
'Chops |
|
| |
|
| Back to top |
|
