MicrobeHunter.com Microscopy Forum

Hi Guys,

i am new here, and looking for solution of my problem. I want to have flat picture of whole surface of cylinder. I mean, i want to rotate cylinder under microscope, and obtain rectangle picture, with lenght equal of circumference... Any idea how to do it?

Thanks,
Simon

Look into image stitching would be my guess.
There's plenty of software, both free and commercial, to do various forms of this sort of image manipulation. There may be better methods suggested by someone who's actually done it, but that's where I would start.

Edit: If you're really dedicated, perhaps some specialized form of photogrammetry might also work too, but I suspect that's a deeper rabbithole than you might be looking for.

I suspect that the combination of a line-scan camera and a stepper motor would be a very good place to start.

https://www.stemmer-imaging.com/en-gb/t ... n-cameras/

MichaelG.

.
P.S __ what are the dimensions of your cylinder ?

bullets, 4,5-13mm

Simon

dtsh wrote: ↑

Tue Mar 02, 2021 3:02 pm

Look into image stitching would be my guess.

What about "no depth of field" with microscope? When i observe cylinder, i have sharp screen only in focused area (belt). So i have to stich tens of slices...

predykat wrote: ↑

Tue Mar 02, 2021 10:33 pm

dtsh wrote: ↑

Tue Mar 02, 2021 3:02 pm

Look into image stitching would be my guess.

What about "no depth of field" with microscope? When i observe cylinder, i have sharp screen only in focused area (belt). So i have to stich tens of slices...

If you're lucky only 10's; that's the cost of fidelity in this regard. The higher the magnification, the less depth of field. Even without the limitations of depth, if your goal is to make an accurate representation of a cylinder the area that would be presented as "flat" will be small in proportion to the entire circumfrence; as the view moves away from the center of the cylinder, the view will become more and more oblique. Depth can also be gained via focus stacking, but the number of images will go up significantly.

How much magnification do you require? Perhaps a stereo microscope would do what you need? They tend to have much greater depth of field due to the typically lower magnifications. If you are looking at a magnification of less than around 70x, that would probably be the better choice.

As for the effort, if this is a one-time type thing, it might be easier to just do it manally, but if this will be repeated, time spent automating (arduino, rpi, etc, etc, etc) will be preferred. Once you're at that point, who cares how many images it is so long as the hardware can process them.

Perhaps a video of the turning cylinder may be stitched to a panorama of the video frames, which would give you what you need? Video-scan stitching is very fast - the software does it all. I use this routinely to image whole specimens on slides rather than take dozens of stills manually. The results are very good and the DOF is larger with lower powered (lower n.a. actually) objectives, and may be increased by closing down the field iris within reason too.. The free and superfast MS-ICE does a great job of video-stitching...

predykat wrote: ↑

Tue Mar 02, 2021 10:33 pm

What about "no depth of field" with microscope? When i observe cylinder, i have sharp screen only in focused area (belt).

.

That’s why I suggested a line-scan

MichaelG.
.
Edit: http://homepages.inf.ed.ac.uk/rbf/CVonl ... nescan.pdf
.
Edit: just found this: https://coptonix.com/en/products/usb-line-camera-8m/

MichaelG. wrote: ↑

Tue Mar 02, 2021 11:37 pm

.
Edit: just found this: https://coptonix.com/en/products/usb-line-camera-8m/

maybe this? https://www.aliexpress.com/item/1005002 ... hweb201603_

predykat wrote: ↑

Wed Mar 03, 2021 7:33 am

maybe this?

Maybe ...

It’s a big market, and [if funds are no problem], I would start by looking at the established suppliers like Basler and Teledyne DALSA

https://www.baslerweb.com/en/products/c ... ht-choice/

https://www.teledynedalsa.com/en/produc ... g/cameras/

Sorry, I don’t know enough to advise you on a purchase ... I just wanted to point you in what I believe is the right technical direction.

MichaelG.

MichaelG. wrote: ↑

Tue Mar 02, 2021 3:39 pm

I suspect that the combination of a line-scan camera and a stepper motor would be a very good place to start.

I don't know about the OP, but thank you for leading me to that rabbithole.

thanks guys.

There may be something of interest to experimenters, here:
https://www.diyphotography.net/emulate- ... rd-images/

Looks like a low-cost way of trying slit-scan

MichaelG.

MichaelG. wrote: ↑

Sat Mar 06, 2021 9:14 pm

trying slit-scan

Thank you for the link.
A whole new hobby opens up for me !

I am old enough to remember school group photos obtained with rotating cameras where inevitably someone had to run from one end to the other to appear twice ( or sometimes not at all when the whole 3-tier stack of pupils on precarious benches were destabilised and ended up in a heap )

All I have to do now is learn this "Processing"-Java environment.

On a slightly disturbing note ...

I attempted to download "Processing" and my Mac warned me that there was a problem with the Security Certificate.

... It is issued by Comodo [allegedly the #1 issuer in the World]

I will investigate further, before proceeding.

MichaelG.

.

UPDATE:
Comodo, and the Processing Foundation both appear to be highly reputable ... so I decided to risk it
Downloaded the latest version of Processing, and followed the instructions on the SlitCam page
https://www.slitcam.com/download.php

The demo kinda works, but I am getting some error messages ... probably relating to the 'version'
More research required ... never easy, is it ?!
Ref. https://github.com/processing/processin ... ges-in-3.0

.

Still struggling with the software, but I thought these might be of interest:

https://www.nist.gov/pml/about-pml/pml- ... s/csi-nist

https://sha.org/documents/Technical_bri ... icle04.pdf

MichaelG.

I have a couple of drum scanners which I use to scan large format film. Much like my Zeiss Universal, these were $50,000 items I picked up on ebay for a couple of hundred. It is an interesting combination of mainly analogue components giving a digital result.

The chassis/ arrangement is not dissimilar to a lathe - the drum is spun by a keylock 'chuck' on the left hand side, then an optical head does the scan one point at a time, which is connected to three (R,G,B) photomultipliers by optical fibres. Unlike a lathe the optical head is static and the drum, chuck and motor move across it along two precision ways made of steel bar. Trans-illumination is provided by optical fibres running inside the acrylic drum, with the film 'oiled' to the drum with scanning fluid (I use a painter's thinner called Gamsol).

The scan is done in a spiral like cutting a screw thread. You can define the dpi which is controlled by sample rate, drum speed in x and y, and choose the aperture. Colour fidelity is excellent as RGB are all sampled at each site and lower res scans scan a larger point rather than simply spacing out the sampling points.

For something small like a bullet I imagine you could mount on a micro-lathe, photograph strips and stitch.