I tried to research a field of 3d scanning technology and became a bit confused with the technology. I’ve read that there are laser scanners and Structured light scanners. But Isn’t laser a structured light? What’s principal difference, and how the light can be structured and not being a laser. For example Faro is laser 3d scanners, while Artec - structured light ones.
The key difference of a scanning laser vs structured light is like comparing analog to digital
The laser beam (line) is recorded via a camera sensor the warp of the line is calculated based on pixel offset from the expected straight line. - basic analog type function
Typically the resolution of the X,Y,Z are all equal and accuracy is limited by the laser linewidth on the object
Structured uses a combination of complex patterns that are projected onto the object and captured by a camera sensor.
With structured light the resolution typically (depends on number of patterns used and algorithm) can be 1/10th of the projected pixel size, Z resolution is typically 10x of the X,Y resolution
Structured light captures all the data in a single capture, with scanning laser needs to scan the object one scan at a time
1 Like
So there really is a lot going on with that question. I’ll help elaborate on some 3d data acquisition systems and their basic function. These have been merged and blended together quite a bit over the past 10-15 years, so some there aren’t as many ‘pure’ systems as there once were. All of these systems work on pretty much the same basic principle, triangulation of scanner to object distance, although get there in slightly different methods.
Structured Light Scanner: This projects pattern(s) of light, typically lines or boxes, which is used to image the relative distance between every visible point from the camera to the sensor. I believe this is done through triangulation on many edges simultaneously. The scanner and subject can’t be moved separate from each other during a single ‘shot’, but can be moved in between ‘shots’; think of it like taking a photograph where if you move the camera while the shutter the image will be blurry. Typically ‘stand alone’ systems, but not always.
Pro: Captures an area per ‘shot’. Doesn’t require good positional information for either the scan subject or the scanning sensor. Can usually image around an object quite well without much setup. Used to require ‘target’ stickers or tracked table, doesn’t seem to any longer.
Cons: Not great accuracy, especially in the Z (to/from sensor) direction.
(Single) Laser Line Scanners: This calculates distance one laser line width at a time. This requires either the scanner or the subject to move for the scan to take place. The faro system tracks the scanner movement with the arm, other systems move the subject, other systems track both.
Pros: Typically more accurate than other technologies, given the same measurement volume. Can upgrade numerous manual/CNC equipment to turn it into a measuring device. Versatile implementation potential.
Cons: Can be slow and expensive. Typically has shorter sensor to subject distance which can be a problem (see zeiss’s laser scanner). Doesn’t pick up color information (typically, some have a color camera as well).
Photogrammetry: This uses just photographs without structured light projection. by taking a lot of images with some but not too much camera movement, the camera path can be calculated as can the scene topography. It’s the most computationally intense and requires the really good imaging conditions, but works really well.
Pros: Cheap! It’s just software at this point, as cellphone cameras or SLRs are quite affordable and will make some good scenes. Awesome for drones, too. Can be very fast with practice.
Cons: Not very accurate. Very processing intensive, can be slow to get results.
(multiple) Laser line scanner: I’ve only seen one of these, which projected a rasterized laser projection. Pretty much worked like a structured light system.
Not Triangulation based systems:
Interferometry / Phase shift (Laser trackers): this shoots a laser out and catches the reflection and measures phase shift difference through the interference pattern produced through a beam-split return reflected signal. The beam is tracked via a head that measures azmuth and altitude by tracking the single point.
Pros: can be really long range (60+ meters) and very accurate for those distances ( 0.001" or so).
Cons: long warm-up time. Some require two operators and may not tolerate an interrupted beam. Usually really expensive.
Focus Stack: Microscopes with Z height information can create a 3d image through capturing a series of images and recording the position where a given area is in focus. Typically a wide open aperture is used to reduce the depth of focus to get better vertical differentiation.
Pros: Fairly inexpensive. Works on tiny subjects. same process gives hyper-focus images
Cons: Typically doesn’t have great Z accuracy/resolution, a sphere would likely come out egg shaped. can be difficult with a manual microscope. Doesn’t give true 3D information, it provides topographical information; each x/y position can have only 1 z location.
Confocal Microscopy: This is a specialized subset of focus stacking. Using a second lens and a spinning disc with a very small hole in the optical stack, it creates the ability to physically block all out-of-focus light. by using small focus adjustments this allows for pretty much unparalleled resolution.
Pros: Super high resolution (down to around 50-100nm or less).
Cons: Really expensive. Super slow. Topographical information, not true 3D.
1 Like