motorized laser vs handheld

[mercedesmorita]

Hello, I am wondering if the cause of the stripes in the scan have to do with the fact that I am moving the laser by hand (and thus passing the laser line many times through the same part of the object). Do you know if using a stepmotor, or a cheaper controller such as a kitchen clock, would reduce these stripes?
I tried everything: changing the laser and changing the camera. The stripes appear in every case (more or less).
I would appreciate your answer, thanks!!!

[WalterMo]

My best experience for laser scanning was to use a stepper motor with gear. It's better to sweep one time very slowly above the surface of the object than quickly multiple times.
Some years ago when structured light scanning wasn't possible with DAVID we have developed several motor scanners. Here are 2 of all:
viewtopic.php?f=6&t=1999

viewtopic.php?f=6&t=2421&p=14642&hilit

viewtopic.php?f=6&t=1827&hilit


But anyways, please post a screen shot of your hand scan. Maybe we can find a solution to the lines.

[mercedesmorita]

Hello Walter, thank you very much for your reply.
Here are my scans (I am comparing the results of the david laser scanner with photogrammetry, so you will see that the image on the left is of photogrammetry):
(Photogrammetry and laser scanning with Thorlabs camera 1645c 1280 x 1024 25fps)
(Photogrammetry and laser scanning with Microsoft cinema webcam 1280x720 10 fps)
(Photogrammetry and laser scanning with Thorlabs camera 1645c 1280 x 1024 25fps)
(Photogrammetry and laser scanning with Microsoft cinema webcam 800x448 15fps)
(Photogrammetry and laser scanning with Microsoft cinema webcam)

As you can see, the results obtained with laser scanning are not as good as the ones obtained with photogrammetry. But I am thinking that my problem maybe is one of these:

-I am using a red laser instead of a green one
-The thickness of my laser line is about 1 mm (the object is 5 cm of height, maybe the line should be of 0.5 mm)
-The fps? or the resolution? I tried different cameras: Thorlabs camera 1645c 1280 x 1024 25fps and a Microsoft cinema webcam 1280x720 10 fps. The results shows the same problem: the lines.
- Or it is like I told you before, the fact that I am passing the line multiple times (not quickly, but manually)

Hope you find the problem. Thanks!!

[WalterMo]

I have made some laser scans to find a solution to your line problems.

My test object was a roll of kitchen towel. A bit pressed flat at the scan side. Laser was a green 5mW focusable one.
The sans were made by a motor scanner and also by hand. Camera was a modified Logitech C615:
viewtopic.php?f=6&t=5863&p=28683&hilit=c616#p28683
Its lens was a CS zoom 6-15mm, 1/3“ with manual iris.

Firstly I scanned by stepping motor at a camera resolution of 800 x 600 @ 30fps. => Dark blue scan. Speed: 95 mm/min.

Secondly at maximal camera resolution of 1920 x 1080 @5fps. => Green scan. Speed: 22 mm/min.

Thirdly by hand at 800 x 600, trying to hold the laser in such a way that its line was horizontally. => Ochre scan. The paper roll was multiple times „swept over“ by the laser, as long as every part was hit.

And finally by hand at 800 x 600 „holding the laser line“ mainly horizontally but also diagonally. => Light blue scan.

So it's difficult to say what the reason is for your lines. Regarding webcams: Best results were made with Logitech devices. Pro9000, C600, C615 and similar. But it's not due to the laser colour. Red is also fine. Green is better to "see" for a webcam. A focusable laser is recommended. 0.5mm is OK. And disable all automatic functions of the camera!

Also important is that the laser and the camera are focused sharp on the scan object's surface and not on the corner walls. Hold the laser line horizontally. If you cannot hit every part, e.g. at your posted stone relief of a human head, you have to turn it and do additional scans. Allign and then fuse them.

[mercedesmorita]

Thank you, Walter.
So maybe the problem is the camera? It is strange, because I am using a Thorlabs camera (we use it for scientific work in our lab):
DCC1645C - High Resolution USB2.0 CMOS Camera, 1280 x 1024, Color Sensor
optical sensor class: 1/3"
pixel clock range: 5 - 40 MHz

I will try doing it again with your reccomandations. So, you are saying that I don`t have to rotate the laser, but only move it in the same line (like in a stepper motor) an well centered?
I will try it and I will show you the results soon.
Thanks again!

Mercedes

[WalterMo]

Yes, don't rotate the laser around its longitudinal axis. Pan the laser up and down when the line is circa horizontally. The so called intersection angle should be about 30°:
viewtopic.php?f=2&t=3117&p=17498&hilit

The camera should see this: On each wall side of the calibration corner ca. ¼ of the laser line and ½ on the object.

Some years ago we have developed the so called plane-less laser scanning. That means that no calibration corner is necessary during the object scan. Only 1x for the reference scan. Advantage: All camera pixels can be used for the object scan. And, no corner has to be behind the object.
But: A precise motor scanner is necessary which can exactly do the very same reference scan and later on the object scans.
But now with structured light it isn't reasonable to use plane-less laser scanning. It's "automatically" plane-less scanning.

Walter

[mercedesmorita]

Hi Walter,

I used a stepper motor and definitely the results are better!

IMG]http://i58.tinypic.com/2ewjnma.jpg[/IMG]

Now I am trying to change the parameters (size of the panels, thickness of the line, frames per second... to understand the cause of these lines and improve these results.

Thanks!

[mercedesmorita]

Sorry, here is the image:

[WalterMo]

Hi Mercedes,

Good to see your better results.
What kind of stepper motor have you used? With or without an additional gear? Micro step mode?

Best would be if you scan something for testing and comparison that we know (as a standard). Maybe also a paper roll.
Or simply scan the corner walls. But for this you have to change one parameter in the Advanced Settings (the little cogwheel symbol top right):
Go to 'Laserscanner', 'Triangulation', change the 'BackgroundFilterFactor' from '0.1' to '-0.1'.

Was the red laser a focusable one?

Walter

[mercedesmorita]

Hi Walter,

I used a micro stepper motor (resolution of 2,5 microns).
The laser was focusable. But I notice (it is logical) that the focus changes as the line moves over the different reliefs (over the parts with more volume the laser line changes, because my object is very volumetric). So, the parts where the line is expanded are less accurate.
I thought that maybe the lines in the scan were caused by this thickening, so I did some tests, as you told me, scanning the panels to study the thickness of the line constantly. And here are my results:


above, with a line of less than 1mm
down, with a blurred, thick line, approximately 3 mm

So, even with the blurred laser line there is no effect of the small horizontal lines in the scan.

On the other hand I have a question:
Changing from 0.1 to -0.1 allows me to scan the panels. Then, I don't understand the function of the calibration. Because I can scan also the object with this -0.1 (although the panels also appear).

Sorry for my bad english

Thanks

Mercedes

[mercedesmorita]

Oh, I forgot to ask you about that "drawing" observed in the last scans (Is it related to a poor quality of the laser line? Maybe the lens of the laser is damaged?

[WalterMo]

Hello Mercedes,
The Background Filter Factor defines a distance from the corner panels where 3D points will be ignored by DAVID. 0.1 means 10% of the calibration grid space (= ¼ of the scale length) is ignored. Any negative value will allow scanning of the panels itselves and all things in front of them. The calibration has nothing to do with the Background Filter Factor. This factor works during scanning. You can change it as you like, the calibration is still valid.
I think the black patterns („drawing“) is due to unwanted laser light reflexions from maybe the glossy table surface. Lay a black piece of cloth between table and calibration corner.
Regarding the depth of sharpness of laser and camera: Focus both devices to an average plane of the object. A small aperture (high value) of the camera lens iris will give best depth of field sharpness.

Walter

PS. A hint: If you work with parameters in the Advances Settings, shift the mouse cursor on the term and you will see (for a moment) their meaning.

Edit. For comparison I have made a motor scan of my calibration corner. Camera: Modified Logitech C615 at 800x600 pixels and @30fps. No interpolation or smoothing.
Laser was above the camera. Speed: 95mm/min. (First screen dump)

Second screen dump: Modified Logitech C600 at 1600x1200 pixels @5fps. Speed 22mm/min. Both camera lenses have a mechanical iris aperture.

[mercedesmorita]

Hi Walter,

Thanks a lot for your advices and explanations.
I will try to improve the scans with all that.
If I have other questions I will write again in the forum.

Regards

Mercedes

[Oteck]

Hello all, i'm still new but i'm wondering if would make any difference if you placed a laser on a linear rail cnc slider so rather than sweeping through a fixed pivot the laser will pan across the scan

[WalterMo]

Hi Oteck,
I don't know what kind of movement gives better results. Fact is that the so-called intersection angle for laser scanning should be about 30-40°. It is the angle between the laser plane and a lot of imaginary lines between the camera and the place where the laser just illuminates the object and the walls left and right. See the sketch here:
viewtopic.php?f=2&t=3117&p=17498#p17498

MagWeb has investigated the behavior of the intersection angle for different laser positions, if the laser is panned or if it is linearly moved, 22 July 2008:
viewtopic.php?f=9&t=588

You see that this angle is paradoxically more constant if the laser is panned.