Re: General Question on Z focus series. [SEC=UNCLASSIFIED]

[Goldsmith, Noel]

Rhod Wilson asks the following.


On 12/2/10 4:00 PM, "IMAGEJ automatic digest system" <[hidden email]>
wrote:

> Date:    Thu, 11 Feb 2010 11:38:29 -0000
> From:    "Wilson R S (AT)" <[hidden email]>
> Subject: General Question
>
> Does anyone know of any information regarding the automation of system
> to create extended depth of focus images?
>
> i.e. Collect images in a stack and pass to a plugin, for example: stack
> focuser or extend depth of focus.
>
> =20
>
> I am trying to create a system that collects images of particles
> suspended within a microscope slide (not all at the same level), so the
> start point for collection needs to found in the automation and not hard
> programmed.
>
> =20
>
> My main issue is locating this start position, from here I have good
> system.
>
> Any help would be much appreciated.
>
> =20
>
> Rhod

Noel Replies.
We routinely produce "Deep-Focus" (it's in quotes because it is strictly
deep field, the image is on the focus side) stacks of images.
So far our approach to starting and stopping the sequence of through focus
images is limited to the following procedure.
1. The user positions a specimen, and makes sure that the features they
require are appropriately framed and the camera will take images which have
a suitable exposure and contrast range (no saturation).
2. A "Live" image is displayed and the user is asked to focus on the
"highest point" (which means that they visually scan the live image and look
for some small region in focus, which is that place which is highest). They
then click the OK button.
3. They are then asked to focus on the lowest point, and they click the OK
button.
The automation then assumes control over the camera and the stage and takes
as many images as are needed to cover the range defined above, taking into
account the depth of field (focus) of the selected objective.
We have considered the option of performing the discovery of the high and
low points in images, and certainly consider it attractive.

The potential drawbacks are.
1. There would need to be a setting, particular to each specimen and lens,
to prevent collisions, this would need to be set up for each specimen in our
case as we look at rough, irregular pieces of broken metal, composites,
polymers and ceramics. To achieve this setting would take about as long as
what we already do.
If you are looking at well defined regular samples you can trust then life
may be simpler. One setting might do all. And as you have slides, I guess
that would be true.

2.Then there is the problem of deciding when an image has anything in focus
or not.
Probably "find edges" would do it. This would mean a loop to take images
(probably of reduced resolution to boost speed), and then run "find edges",
look at the amount of edge in the result and repeat until you get some. Then
move to the other extreme and repeat. Then add a small Z buffer to each end
and take the Z series. And in your case, just finding the "start" position
is enough. So you would only have to this once per series.

It seems to me that a good reason to do this automatically would be that it
costs more to do it manually.
So I will be thinking of ways to perform this automatically, but see that it
is actually quite a difficult problem.
I know this hasn't solved your problem, but I hope that you find this
helpful.
Regards
Noel

Noel Goldsmith
Air Vehicles Division
DSTO
506 Lorimer Street
Port Melbourne
3207 Victoria
AUSTRALIA
613 96267527
0428364003
[hidden email]
--



IMPORTANT: This email remains the property of the Australian Defence Organisation and is subject to the jurisdiction of section 70 of the CRIMES ACT 1914.  If you have received this email in error, you are requested to contact the sender and delete the email.