NTNU – Application of 3D Machine Vision (TPK4850)

In this master course “experts in team” the scope is defined to:

“investigate new strategies, techniques and management, which lead to development of ecological and practical 3D machine vision systems in a total deferent way than traditional ones”

Read more @ NTNU: Application-of-3d-machine-vision-tpk4850

The course describes the following Scorpion 3D Stinger based system for

3D pizza sorting system

3D Pizza sorting system
The modular system comprises a conveyor, an ejector unit, machine vision hardware and software. The system is designed to meet the ip-grades and wash down requirements of the food industry. The following core functions describe the system:  Locate the pizza, measure the diameter, thickness profile and edge defects. Based on the measurements the user can set up the sorting criteria in a user friendly graphical user interface. The capacity of an installed system is 7200 pizzas per hour at a conveyor speed of 0.5 – 1.0 m/s. The system handles multiple and different sizes through a flexible recipe system. The diameter resolution is 0.5 mm and the height resolution is 0.2 mm.

3d image data

The 3D image data is used to accurately measure the height of the products in different regions of the surface. Local minima and maxima are localised.

The machine is based on the same powerful Scorpion Vision Software that has been at the heart of many automatic inspection systems used in manufacturing industries. Intended to reduce wastage and improve quality, the system uses a series of cameras to build a 3D profile of each product.

3D bin picking solution

Norwegian R&D project AutoCast developes leading edge 3D bin picking solution for the casting industry.

3D bin picking solution for the casting industry.

The leading Norwegian manufactures of castings need new technology to meet the international competition and the environment-, health- and safety requirements. This implies a considerable emphasis on the development of new technology to enhance the use of automation and robotics in the foundries. The small series production does not allow for unique solutions for individual parts. The main challenge is therefore to develop generic solutions that can be used on several different parts or families of parts.

Considerable focus on R & D has put the foundry industry in a leading position regarding casting- and materials technology. However, high salary costs and lack of working resources has created a need for further development of technology within automation and robotics. The Norwegian foundry industry has put their resources together through a R&D project, called “AutoCast” in order to focus on this development. The goal is to develop state of the art technology that ensures competitiveness of the Norwegian foundry industry.

Autocast project

AutoCast started in 2008 and is going to run for four years. The total budget is 5 MEuro. The project is supported by the Norwegian Research Council.

Tordivel AS, the company behind Scorpion Vision Software, is chosen as the 3D Machine Vision Partner for the AutoCast project.

In 2005 Tordivel decided to move Scorpion from a robust and proven 3D platform. The development has been organized in several R&D projects. The EU-CRAFT consortium 3DMulticam has developed a platform for 3D underwater metrology while the Auto3D project sponsored by the Norwegian Research Council has focused on creating a low-cost complete framework for 3D Machine Vision including stereo-vision, stripe light, 3D visualization, 3D references, doing 3D in 2D images and true 3D processing in point cloud.


The following scenarios must be handled in the AutoCast project:

  • Identification and location on a conveyor
  • Segmentation and 6 DOF location in a Bin

To solve the 3D Bin Picking solution it was obvious that a 3D machine vision solution was needed. After evaluating stripe light, 3D laser scanners and stereo vision, a solution based on three GigE cameras and 3D stereo vision was selected.

The reason for selecting this solution was the following:

  • Fast – a part is located in less than 1 second
  • Works even if objects are moving on a conveyor
  • Low component cost
  • Robust and reliable with 3 cameras
  • Accuracy

The 3D machine vision technology developed in the Norwegian AutoCast R&D project will keep the Norwegian foundry industry in a leading position regarding casting and material handling. The 3D machine vision part is based on Scorpion Vision Software, off-the-shelf software from Tordivel AS. This concept for accurate, low-cost 3D Bin Picking can be used in a vast number of industries



Scorpion 3D Stinger for Robot Vision


Scorpion 3D Stinger™ for Robot Vision is designed to solve manufacturers’ classic challenge:

  • Picking parts from a conveyor belt,
  • a pallet or
  • a crate.

Scorpion 3D Stinger™ captures images, identifies and locates the product and sends the id and 3D location to a picking robot.

Scorpion 3D Stinger wins the Vision System Design Innovator Award 2016.

Read more:

The Scorpion 3D Stinger™ Camera is designed to be used in cutting-edge 3D Stereo vision applications.

It is compatible with the latest 3D version of Scorpion Vision Software and locates moving objects in 3D in real-time.

Video – 3D Vision Magic – Box Picking and Identification

is-2015-0005-D 3D ImageThis video presents the kind of 3D Magic we like at Tordivel based on the latest Scorpion Vision XI software and the new industrial Scorpion 3D Stinger 30 mm baseline Camera.

The video shows how to locate and identify boxes with a Scorpion 3D Stinger 30 mm baseline Camera. Multiple boxes are located and sorted in 3D space. The Scorpion 3D Stinger Technology is ready to use in

  • 3D Robot Vision
  • 3D Assembly Verification
  • 3D Measurements

The Scorpion 3D Stinger Technology has been developed in Norway since 2000 in a continous R&D effort sponsored by the Norwegian Government and EU.


Read more about:

Scorpion Vision Version XI is here!

Scorpion Download X.I

Scorpion Vision XI Download

We are proud to announce that Scorpion Version XI build 572 is available. The release contains the aggregated updates since Scorpion Version X was released in October 2012 more than two years ago.

This major release of Scorpion Vision Software has many new features and possibilities.

  • Compatible with latest Scorpion 2D and 3D Stinger hardware
  • Faster image processing and more tools supporting multi-core processing
    • Many tools has prefiltering option
  • Improved OpenCV 2.4.9 and Python 2.7 support
    • The move to Pythons improves the ability to use 3rd party libraries
  • Updates Scorpion Tool Components – STC
    • Components for 3D Calibration, dense 3D image generation and more
    • Updated image filtering components
    • Support for 3D Factory Calibration
  • Improved Scorpion Camera Drivers for Sony and Basler GigE Cameras
    • First version of  port based making it easier to switch cameras.

The release will support the upcoming OpenCV 3.0 when released in 2015.

Read more about upgrading to Scorpion Vision XI

Prefilter – a new great feature in Scorpion Vision

The upcoming release includes a new prefilter feature for many tools, these have been added in the recent months. The last tool to be updated is the important feature is the 2D Calibrator tool. This tools provides the means and the a requirement to get sub-pixel accuracy in machine vision – this is important for 2D but also the foundation for the Scorpion 3D Stinger Technology.

The standard calibrator with no pre-filter applied:

The image below shows the calibrator with no pre-filter applied. The calibration image is difficult to use because of low contrast. The calibrator quality measure says that the RMS is 0.180 pixels or 1 / 5.5 pixels.

Scorpion Vision's Calibrator with no prefilter

Scorpion Vision’s Calibrator with no prefilter

Applying a Normalising filter and a small blur

The Prefilter is active with the following command: “N101,255b3”. This means that Image Normalisation with a block size of 101 is first applied the result image is then blurred with a block size of 3.

The important effect to the image is the following:

  • Constrast is much better
  • All image intensity skew or variations are removed
  • The thresholding of the markers are now easy
  • The accuracy is marginally improved – also important

Read more about the available pre-filter options: ImageFilters.htm

Calibrator tool with prefilter active

Calibrator tool with prefilter active

Changing polynominal order from 5 to 9 – improved the accuracy by 60% 

The RMS goes from 0.172 to 0.110 – this means that the optical resolution goes from 1/6 th of a pixel to 1 /9 th of a pixel or 110 / 172 or 60% by changing a single number from 5 to 9. 

In a 3D robot vision system you can hear this when the robot is picking 🙂

The polynominal order is very important - almost double the accuracy

The polynominal order is very important – almost doubles the accuracy

Video – Realtime 3D Imaging of Box Picking

The Scorpion 3D Stinger RPP Camera provides the basis for robust and reliable robot vision solutions. This video deomstrates the camera in a box picking concept showing a simulated robot picking sequence using the camera for live realtime 3D image capture.

The Scorpion 3D Stinger Camera is designed to be used in cutting-edge 3D Stereo Vision applications with or without structured light. It is compatible with the latest 3D version of Scorpion Vision Software and locates moving objects in 3D in real-time.

Application areas
The unit is designed to be used in 3D Robot Vision, 3D Laser Triangulation, 3D Gauging, 3D Assembly Verification and other advanced Machine Vision Solutions. The components are selected with the highest quality. The unit’s design will save man-hours when creating and deploying 3D Machine Vision Solutions. The standard unit, with a nominal weight of 3.5 kg, can be mounted on a robot.

News – 3D Stinger creates 3D Images in real-time for Robot Vision

Scorpion 3D Stinger Laser Camera

Scorpion 3D Stinger Laser Camera


The new Scorpion 3D Stinger RPP Laser Camera is  based on the principle “projected texture stereo vision” to create a high quality dense point cloud.  The 3D camera works on moving objects with or without texture.

“We are proud to present this exciting new option in the third generation Scorpion 3D Stinger Camera first introduced in 2011.  We have perfected the unit by adding advanced integrated laser and led control. The advanced control allows us to capture both 3D images and 2D images within a compact and robust unit.  An important advantage is the unique ability to measure 3D in the 2D images using high precision 3D references. The  unit extends the complete Scorpion Compact Vision product range that provides state-of-art PC based machine vision with low deployment cost.” – Thor Vollset – Founder and CEO of Tordivel AS

Application areas:

Robotics, 3D robot vision, logistics, empty verification, assembly verification, measurement, R&D and more. 

Scorpion 3D Stinger Camera:

The Scorpion 3D Stinger Camera is designed to be used in cutting-edge 3D Stereo Vision applications with or without structured light. It is built with only the highest quality components to be used in professional machine vision with 24-7  operation. It is compatible with the latest 3D versions of Scorpion Vision Software and locates moving objects in 3D in real-time.

Read more about Scorpion 3D Stinger 


The Scorpion 3D Stinger™ RPP Laser Camera consists of:

  • IP-64 aluminium enclosure – typical 3.5 kg
    • to be mounted above the object or mounted on the robot
    • Black Anodized Aluminium
    • Size: L=292 mm x D=266 mm x H=100  mm
  • Two XCG GigE Cameras from Sony with lenses with resolution from VGA to 5MP
    • Baseline of 200 mm – allows camera tilting to handle multiple operation distances
  • High power white or Infrared strobed LED
  • IR – 830 nm and Red – 660 nm Random Pattern Projection laser from Osela.
  • Termination panel & cables
  • Scorpion Stinger Control Card to driver LED and Laser and to control the image capture an strobing sequence
  • 24v input, 5v and 12v output – Power consumption from 15 to 35 W
  • Working Distances from 400 – 5000 mm
  • GigE switch

The control card provides the ability to capture one 3D image set illuminated with a laser pattern and then capturing a high resolution 2D image set.  The 3D images are used to locate objects in 3D. In the 2D image, the 3D references from the 3D image is used to do 3D measurements in 2D images with the highest possible accuracy.

Read more in the Scorpion 3D Stinger Camera datasheet

Principle of operation:

The new camera is based on the third generation Scorpion 3D Stinger Camera where support for multiple LED and Laser Source are controlled by the Scorpion 3D Stinger Control Card. The goal of the project was to develope a high precision and robust  3D Image or dense point cloud systems with an active Random Pattern Projection to be able to work with all surfaces. We uses two 3D Stereo Vision Camera and a Random Pattern Projection Laser to create a high quality dense point cloud.

The input:

  • A 3D image set with projected texture from a RPP laser
  • A 2D image set illuminated by high power LED
Scorpion 3D Stinger RPP Laser Principle

Scorpion 3D Stinger RPP Laser Principle

The result:

  • A dense 3D point cloud
  • A 3D height map

Supporting universities, students and colleges

We supports student doing their machine vision and robot vision projects and universities that are in the process of building labs for 3D Robot Vision or 3D machine vision. We can tailor sw and hw packages to any need both for reasearch and professional use.