Google Groups no longer supports new Usenet posts or subscriptions. Historical content remains viewable.
Dismiss
Application/library to analyze photos?
16 views
Skip to first unread message
Gilles
May 17, 2012, 7:39:44 AM5/17/12
to
Hello
A friend who works in the medical profession asked me to check if an
application or library were available to analyze pictures of wounds
taken at the hospital (length+width+depth, colors).
IOW, we're talking handheld, multiple shots over a period of
days/weeks, light/distance a bit different each time: I know nothing
about image processing, but I assume it must be pretty hard to do and
if such application does exist, it's a tiny market and sold at a
premium.
Ideally, it should run on smartphones so that it's self-contained, ie.
not dependent on some remote, fast PC for image processing.
I read about OpenCV, but can't tell if it's good enough to analyze
such pictures.
Any infos on applications/libraries that can do this, even for
Windows/Macs.
Thank you.
A friend who works in the medical profession asked me to check if an
application or library were available to analyze pictures of wounds
taken at the hospital (length+width+depth, colors).
IOW, we're talking handheld, multiple shots over a period of
days/weeks, light/distance a bit different each time: I know nothing
about image processing, but I assume it must be pretty hard to do and
if such application does exist, it's a tiny market and sold at a
premium.
Ideally, it should run on smartphones so that it's self-contained, ie.
not dependent on some remote, fast PC for image processing.
I read about OpenCV, but can't tell if it's good enough to analyze
such pictures.
Any infos on applications/libraries that can do this, even for
Windows/Macs.
Thank you.
PiLS
May 17, 2012, 2:43:18 PM5/17/12
to
Gilles <nos...@nospam.com> à ploppé:
I believe ImageJ could be made to do that; it's written in Java so
it should run on anything that can run Java. Depending on the size
of the images, it could be too heavy to run on a handheld, but these
tiny things tend to have impressive specs these days.
It's a bit of a niche application, you're right.
It's available absolutely free of charge at http://rsbweb.nih.gov/ij/
They have a lot of plugin for specific applications, too.
The width, length and area are going to be no problem at all if you
have an object of known size on each image to calibrate the
measurement.
Not sure how you intend to calculate the depth unless you have 3-D
images. If you have then it's a doodle; otherwise you'll have to make
with an area measurement and/or colour analysis. There is no easy way
to reliably calculate the depth from a 2-D image.
(unless you have 2 objects of known size, one at the lip and one at the
deepest point of the wound, then it's simple perspective calculation,
but that implies shoving some stuff into the wound, not sure it's good;
and it's going to be very imprecise unless the wound is much deeper than
the size of the reference objects).
--
PiLS
it should run on anything that can run Java. Depending on the size
of the images, it could be too heavy to run on a handheld, but these
tiny things tend to have impressive specs these days.
It's a bit of a niche application, you're right.
It's available absolutely free of charge at http://rsbweb.nih.gov/ij/
They have a lot of plugin for specific applications, too.
The width, length and area are going to be no problem at all if you
have an object of known size on each image to calibrate the
measurement.
Not sure how you intend to calculate the depth unless you have 3-D
images. If you have then it's a doodle; otherwise you'll have to make
with an area measurement and/or colour analysis. There is no easy way
to reliably calculate the depth from a 2-D image.
(unless you have 2 objects of known size, one at the lip and one at the
deepest point of the wound, then it's simple perspective calculation,
but that implies shoving some stuff into the wound, not sure it's good;
and it's going to be very imprecise unless the wound is much deeper than
the size of the reference objects).
--
PiLS
Gilles
May 17, 2012, 6:30:30 PM5/17/12
to
On Thu, 17 May 2012 18:43:18 +0000 (UTC), PiLS <pi...@invalid.ca>
wrote:
kind of metered square around the wound will be enough for the
application to measure it. That leaves analyzing its aspect to
determine its type and nurses should do.
Thank you.
wrote:
>I believe ImageJ could be made to do that; it's written in Java so
>it should run on anything that can run Java. Depending on the size
>of the images, it could be too heavy to run on a handheld, but these
>tiny things tend to have impressive specs these days.
>
>It's a bit of a niche application, you're right.
>It's available absolutely free of charge at http://rsbweb.nih.gov/ij/
>They have a lot of plugin for specific applications, too.
>
>The width, length and area are going to be no problem at all if you
>have an object of known size on each image to calibrate the
>measurement.
Thanks much for the infos. I'll check ImageJ. Maybe just using some
>it should run on anything that can run Java. Depending on the size
>of the images, it could be too heavy to run on a handheld, but these
>tiny things tend to have impressive specs these days.
>
>It's a bit of a niche application, you're right.
>It's available absolutely free of charge at http://rsbweb.nih.gov/ij/
>They have a lot of plugin for specific applications, too.
>
>The width, length and area are going to be no problem at all if you
>have an object of known size on each image to calibrate the
>measurement.
kind of metered square around the wound will be enough for the
application to measure it. That leaves analyzing its aspect to
determine its type and nurses should do.
Thank you.
PiLS
May 18, 2012, 1:36:25 PM5/18/12
to
Gilles <nos...@nospam.com> ￜ ploppￜ:
ImageJ will give you measurements in pixels (the hotkey is CTRL-M),
but if you calibrate the image using an object of known dimensions it
will give you the measurements in whatever unit you calibrated it in.
--
PiLS
but if you calibrate the image using an object of known dimensions it
will give you the measurements in whatever unit you calibrated it in.
--
PiLS
Gilles
May 22, 2012, 7:14:17 PM5/22/12
to
On Fri, 18 May 2012 17:36:25 +0000 (UTC), PiLS <pi...@invalid.ca>
wrote:
wrote:
>ImageJ will give you measurements in pixels (the hotkey is CTRL-M),
>but if you calibrate the image using an object of known dimensions it
>will give you the measurements in whatever unit you calibrated it in.
Thank you.
>but if you calibrate the image using an object of known dimensions it
>will give you the measurements in whatever unit you calibrated it in.
Gilles
May 23, 2012, 7:31:11 PM5/23/12
to
On Fri, 18 May 2012 17:36:25 +0000 (UTC), PiLS <pi...@invalid.ca>
wrote:
wrote:
>ImageJ will give you measurements in pixels (the hotkey is CTRL-M),
>but if you calibrate the image using an object of known dimensions it
>will give you the measurements in whatever unit you calibrated it in.
Thank you.
>but if you calibrate the image using an object of known dimensions it
>will give you the measurements in whatever unit you calibrated it in.
0 new messages