3D sensing methods compared: which one suits freight?
Time-of-flight, structured light, stereo and laser line scanners at a glance
Request a demoIf you measure freight automatically, you need a three-dimensional image of pallets, packages or long goods. Five optical methods provide that image. They differ mainly in where the depth comes from: from the travel time of light or from an angle between light source and camera.
For freight dimensioning, what counts is how a method copes with what happens in the warehouse: white film, plain-coloured boxes, edges and overhangs, several sensors over the same freight. MetriXFreight measures with time-of-flight. Here is why.
Travel time or angle
A time-of-flight camera measures for each pixel how long its light takes to reach the object and come back. Transmitter and receiver sit in the same place.
Structured light, stereo and laser line scanners work with triangulation: two viewpoints a distance apart see the same point at different angles, and the depth follows from that. This distance, the baseline, determines the accuracy. It can also mean that behind edges there are areas that only one of the two viewpoints sees. With all three methods, the measurement error also grows with the square of the distance.
Five methods, one comparison
Choose what applies in your warehouse. Each card shows the principle and whether the method fits. Below it, all criteria are listed at a glance.
-
Time-of-flight
Travel time of light
The camera emits modulated infrared light and measures for every pixel how long the light takes to reach the object and come back.
A depth image of the whole scene per capture, with no offset between emitter and camera.
Fits Plain surfaces without texture: Depth comes from the travel time of light, not from image features. -
Structured light
Triangulation
A projector casts a known pattern onto the object and a camera offset to the side sees how the pattern deforms. Depth follows from the shift.
High resolution at short range, for example when scanning parts.
Fits Plain surfaces without texture: The projected pattern supplies the features itself. -
Passive stereo
Triangulation
Two cameras set apart capture the scene at the same time. The software finds the same features in both images and computes depth from their offset.
Needs no light source of its own, but needs surfaces with texture.
Does not fit Plain surfaces without texture: On plain surfaces the software finds no features it can match in both images. -
Active stereo
Triangulation
Like passive stereo, plus an infrared projector that casts a dot pattern so that plain surfaces have features too.
Fills the gaps of passive stereo on plain surfaces.
Fits Plain surfaces without texture: The dot pattern gives plain surfaces features too. -
Laser line scanner
Triangulation
A laser draws a line on the object and a camera at an angle measures its shape. The 3D image is built profile by profile while the object or the scanner moves.
Established on conveyors where goods pass through anyway.
Fits Plain surfaces without texture: The laser line itself is the feature the camera detects.
All criteria at a glance
| Criterion | Time-of-flight | Structured light | Passive stereo | Active stereo | Laser line scanner |
|---|---|---|---|---|---|
| Plain surfaces without texture | good Depth comes from the travel time of light, not from image features. | good The projected pattern supplies the features itself. | weak On plain surfaces the software finds no features it can match in both images. | good The dot pattern gives plain surfaces features too. | good The laser line itself is the feature the camera detects. |
| Shadowing at edges | good Without triangulation there are hardly any areas the light cannot reach. | weak Projector and camera look from different angles; depth is missing at edges. | limited Next to edges lie areas only one of the two cameras sees; the larger their spacing, the more. | limited Same geometry as passive stereo: only one camera sees areas next to edges. | limited The larger the angle between laser and camera, the more occlusion. |
| Error as distance grows | limited Noise rises as less light returns with distance. | weak Depth error grows with the square of the distance. | weak Depth error grows with the square of the distance and falls with the camera spacing. | weak Same geometry as passive stereo. | weak Triangulation: the error grows with the square of the distance. |
| Whole scene in one capture | good The whole scene is lit and captured at the same time. | limited Depends on the pattern: one pattern is enough, several patterns need several captures in a row. | good Both cameras trigger at the same time. | good Both cameras trigger at the same time; the dot pattern is always on. | weak Needs the object or the scanner to move to build a 3D image from lines. |
| Object in motion | limited The phase images are taken one after another; motion distorts values at edges. | limited Methods with several patterns distort with motion; single-pattern methods cope with it. | good Works with motion when both cameras trigger in sync. | good Fast methods also suit moving objects. | good Built for it: the motion creates the image. |
| Several sensors on the same freight | limited The cameras interfere, but can be decoupled with different modulation frequencies. | limited Overlapping patterns interfere; countermeasures take effort. | good Passive cameras emit no light and so do not disturb each other. | not rated | not rated |
| Dark, shiny, transparent surfaces Applies to every optical method | weak Little or deflected light comes back. | weak Reflections and transparent surfaces disturb the pattern. | weak Uniform and poorly lit surfaces give no reliable features. | weak On dark surfaces the pattern becomes weak. | weak Dark surfaces return less light, gloss saturates the receiver. |
good · limited · weak · – not rated
How MetriXFreight deals with the limits of time-of-flight
- Motion: MetriXFreight deliberately measures the freight during a brief stop under the system. When this is faster in operation than measuring on the move is shown in Measure standing or driving through.
- Several sensors: time-of-flight cameras disturb each other when they light the same freight. Our systems are built for this and operate their sensors so that they do not disturb each other. That is how the L33x sees the pallet with 3 sensors from all around, and the L63x sees long goods with 6.
What applies to all optical methods
Dark, reflective and transparent surfaces are difficult for every method that measures with light: black foil absorbs the light, metal deflects it, glass lets it through. That is why every data sheet lists limits here. For MetriXFreight, this is how it looks:
- L33x: also measures black foil.
- L23x and L63x: restrictions possible with black foil and transparent or highly reflective surfaces.
- L13x: restrictions possible with transparent or highly reflective surfaces.
- S140: measures all items except entirely non-reflective or highly reflective ones.
Our systems are designed for warehouses without direct sunlight on the measuring surface.
Why metrilus relies on time-of-flight
For freight in the warehouse, time-of-flight brings three properties that matter:
- No structure needed. The depth comes from the travel time of light, not from image features. White film and plain-coloured boxes measure just like printed ones.
- Hardly any shadowing. Without triangulation, there are hardly any areas behind edges that the sensor cannot see. That matters with stacked freight and overhangs.
- No feed motion needed. Unlike a line scanner, the camera captures the whole scene without the freight or the sensor moving. The freight stops briefly under the system, and the sensor captures it completely.
For conveyor belts, where the goods pass through without stopping anyway, laser line scanners are an established method. For pallets, oversize items and long goods that stop briefly under the system to be measured, time-of-flight plays to its strengths.
How a time-of-flight camera measures in detail is explained on the page about the time-of-flight camera.
Frequently asked questions about 3D sensing methods
What is the difference between time-of-flight and structured light?
Time-of-flight measures the travel time of light for each pixel, and transmitter and camera sit in the same place. Structured light projects a pattern and calculates the depth from the angle at which a laterally offset camera sees the pattern. As a result, structured light is more likely to lack depth values at edges, and the error grows with the square of the distance.
Why does passive stereo have problems with white film?
Passive stereo looks for the same features in two camera images. On plain surfaces such as white film there are none, so the software finds no match and no depth. Active stereo therefore also projects a dot pattern.
Do several 3D cameras disturb each other?
All methods with their own light source can disturb each other if they illuminate the same scene. With time-of-flight, the cameras can be decoupled, for example, through different modulation frequencies. MetriXFreight systems operate their sensors so that they do not disturb each other. Passive stereo emits no light of its own and so does not disturb other sensors.
Which 3D sensing method does MetriXFreight use?
All MetriXFreight systems measure with time-of-flight cameras and add colour cameras for photos of the shipment.
Does a time-of-flight system also measure black objects?
Dark surfaces reflect little light and are difficult for all optical methods. With MetriXFreight, the L33x also measures black foil; with the L23x and L63x, restrictions with black foil are possible.