Laser Grid (Or Other Localization System)

[Marked One]

Hello All!

I am working on a scanning experiment, and had a question about using lasers to find the location of the specimen inside the scanning head. I was planning to use two lasers, which would project a laser grid onto both sides of the sample, from both sides of the main optical train:
____________
|->SAMPLE<-|

I would wish to visualize this grid with two webcams, which could then track relative specimen position to scanner rotation and elevation.

First question: is this idea valid? Is there some easier way to find irregular specimens to milimeter (or less) precision with similar cost (as opposed to simply using a LEAP motion?)

Second question: how would one make this grid, which will ideally be regular and in the range of millimeters? Would shining a line laser through a diffraction grating produce this effect, or would I have to use some sort of XY scanning galvanometer arrangement piped through a fiber?

Any ideas would be much appreciated!

Kind Regards,
Ivan Kozlov

 

[Bionic-Badger]

If you've only got two axises, and the specimen doesn't move during scanning, you might be better off attaching a laser/detector pair on a carriage/rail and scanning across the surface of the area you're detecting. Then when your laser beam is interrupted, you record the position. You'd do that for each axis. It could look like this:

A grid could work, but you'd need a whole lot of lasers and detectors. Getting all of them set up, especially at millimeter spacing, would pretty difficult and require a lot of wiring. You'd probably be better off just using the webcam and some image processing methods to detect where the object is. You could even just use the laser to light up the location of the object to make it easier for the camera to pick up.

You could also look into software that does something like what you're requiring, such as David Laserscan which kind of does what you want. It costs money, but might do what you need and may be covered by a research grant or whatever you're using to fund this.

 

[Marked One]

Thank you!

The first idea would certainly work on a stationary object, but unfortunately since the scanner could be used on living specimens we will need some sort of way to constantly keep track of the position of the specimen (This is going to be compounded by the fact that we will be scanning neural tissue for activation patterns).

I was thinking of perhaps using a laser cross to provide suitable contrast, and then training a simple AI/IR Engine to recognize continuous regions of motion and track them?

On a side note, what would be the effect of shining a laser line or cross through a simple diffraction grating? Am I correct in thinking that the latter would produce a distorted grid?

Thank you for the website!

 

[Bionic-Badger]

Assuming your environment doesn't change, I'd just use a camera and image processing. That way you're not disturbing the environment with the laser lights. You can use a temporal difference filter to determine what has changed frame-to-frame.

It doesn't seem as glamorous as using a laser grid, but you'll probably get a lot more accuracy out of the camera method. You could even use a light at an angle to increase shadow contrast. If your specimen reflects certain colors, you could filter out the background too with a filter on the lens. With digital (web) cameras able to film at full speed in 1920x1080 resolution, that is more than enough to film at 1mm resolution for a < 1x1m^2 region. It'd really be a the way to go, and a good way to learn how to use image processing techniques.

A diffraction grating probably won't help much either. It may project some dots on the ground, but unless your subject can really deflect those dots in such a way that it is easy to detect, you might as well just detect it directly with a camera and lights. A distorted grid would actually be a pain to have to interpret using a sensor too.