Welcome

This blog is intended as a resource site for those wanting to learn how to use the Leica BLK360 laser scanner … which also includes me.   There is already a large amount of information on this machine online but it’s ‘disparate’ – what I am trying to do here is pull together the information, specifications, instructions and anecdotal observations together in to one readable and highly graphic blog.

To get you interested why not have a look at this 3 minute video that shows my first multi-scan project – the amazing outside staircase at the McManus Gallery in Dundee, Scotland – 150 years old in 2017.

mcmanus_stair_small

The scan shown in the video has not been tidied up in any way – I will cover that in subsequent posts, nor has it been converted to a mesh – more on that process later too.

I learned a heck of a lot in the 3 hours it took to record these scans – the 3 main ones:

  • buy a power-brick for the iPad.
  • buy or make some heavy white targets to help with registration – heavy so they don’t blow away like my bits of paper did.
  • take somebody with you to keep an eye on gear and let you go to the loo if you need to.

What is the BLK360?

This is a compact, easy to use laser scanner device which can capture coloured 3d point cloud data and comes with a great ‘bespoke’ version of Autodesk’s Recap Pro software (running on the iPad Pro) which allows you to register (join together/overlap) your scans in the field.

What is Laser Scanning

I am not able to fully explain the amazingly complex science behind the electronics in side this scanner device but I can have a go at explaining the principle.

Here’s a thumbnail graphic showing what the scanner does – this is very very slow motion!

ani_02_fast_3

The laser is beamed at a 45 degree mirror which is rotating, this fires out the beam at 90 degrees in an arc.   The arc can only manage 300 degrees as the unit itself gets in the way.

The beam of light is spinning in the vertical plane as you can see above but it is also rotating in the horizontal plane thus generating a spherical projection of beams (click here to watch another mini video).   The graphic below tries to indicate this:

D:LASER SCANNINGBLK360_BLoG3d_visualsscanning passes Layout1

If you multiply the numbers of ‘vertical’ scans (1) by the number of horizontal rotation steps (2) then you get a massive 91 million points – this is probably considerably more than are actually recorded in a scan.   I have taken the Leica specification ‘accuracy level’ of 6mm accuracy at 10m in to my calculation here.

A laser beam will do one of two things … it will hit something or it won’t.   The latter beams are lost and do not reflect back.   Those beams that hit something return to the device, hit the 45 degree mirror and re-enter the unit to be ‘collected’ and ‘scrutinised’ by two types of sensors.

One measures the time it took to return, the ‘time of flight’, and the other analyses the wavelength that returns which determines the colour of the object that the beam hit.

This all happens at light speed – it really is mind blowing science and the accuracy of the electronic engineering is staggering.   You have to just put that to one side and get on with enjoying the device and marvel at how useful it is for capturing accurate site information very quickly.

Joining Scans Together – ‘Registering’

You can’t collect all the information for a site in one single scan, the scanner unit is moved to get a better view and reduce the amount of the scene that is ‘occluded’.

The more projections in all 3 directions (x, y and z) the better the ‘balance’ is for your registration and the more accurate your overall project will be.

registration_mini

Below is a good description of how registration works – this text is by Marc Zinck – a Contributor to the Recap Pro Forum.

The effect of Balance on the registration

Ensuring sufficient balance in registration is critical to achieving high quality registration results. Balance is a metric which encodes the relative contribution of the data in three orthonormal directions, that is to say, it indicates how ‘constrained’ the registration is in different directions.

Imagine that you have two scans from the middle of an empty lot. The dominant feature is the ground, and all points exist in a single plane. This configuration constrains the registration strongly in the vertical direction, however, since there is a lack of additional features, it is not possible to accurately lock down the relative horizontal position of the two scans.

Now imagine a long flat brick wall in the middle of this otherwise empty lot. That wall will constrain the registration in a second dimension, specifically the direction normal to the wall, however the registration is still not fully constrained since it is still not possible to accurately determine where along the wall to place the scans. It may help to think of the scans being able to “slide” along one another.

Finally, imagine a second wall, perpendicular to the first wall. At this point there are strong features in three perpendicular, or orthonormal, directions. Scans which contain all three features (the ground and the two walls), should register together with good values for balance.

Real world environments where this issue can arise include long and straight featureless corridors.

I encourage you to pay attention to the balance metric, it may be able to help guide your scan station placement such that they contain enough features in different directions to achieve high quality registration results.

Another succinct description of how Recap Pro determines how accuracy your scan registration is:

registration_result

Overlap represents the amount of scan data that can be used for registration. ReCap Pro recommends at least 30% of overlap between each scan in order to process the registration properly.

Balance represents the quality of the features (or surfaces) used for registration. Features with surfaces that face in all x, y and z directions will produce better results than surfaces that face in one or two directions.

The points less than 6 millimeters inquiry section tracks how many points in the unregistered scan are within 6mm of the registered data. Data in this range indicate a consistent match between scans.

Modelling from scans – http://constructible.trimble.com/mep-industry/how-to-create-a-3d-cad-model-from-a-laser-scanned-point-cloud-overview.

registration_tips