How to get the best 3D stereo camera calibration with SceneScan/ SceneScan Pro?

Although we send our stereo vision system pre-calibrated and auto re-calibration is running constantly, you will not be able to avoid one or the other calibration with your system. A manual calibration is good, because you can be sure about maintaining the best measurement performances.

Nevertheless calibration should not consume too much of your precious time, nor cause trouble. Therefore we have written this guideline to show some tips and tricks to get the system calibrated and running again as quickly as possible.

The calibration procedure at a glance

For those in a hurry, here is the calibration procedure at a glace:

  1. Go to the “calibrate camera” page of the SceneScan web interface
  2. Place the Siemens star at the desired measurement distance
  3. Make sure that the camera apertures are fully open
  4. Adjust the focus on the Siemens star at the camera lenses
  5. Fasten the focus lock screws
  6. Check if the correct calibration pattern size is selected
  7. Chose between auto capture or single frame
  8. Make sure to detect the calibration pattern from as near to far as possible in every critical corner, detecting the pattern in both field of views of the cameras
  9. Take at least 20 frames for calibration
  10. After calibration, adjust the lenses to the desired aperture if needed.

The right focus for best image quality

The correct focus is essential for optimum image quality of the measurement results. The desired measuring distance should be reproduced as sharp as possible in order to guarantee pixel accuracy of each measuring object and its edges.
So if you want to measure objects at a minimum distance of 1m, you have to focus on 1m using the Siemens star. Setting a precise focus is easiest when the aperture is fully open (smallest f-number). The desired aperture can be set after the focus has been adjusted.

Consider that each change of the minimum measurement distance means changing the focus and thus requires a re-calibration of the system.

Working with the calibration pattern

Tip 1:

Unlike focusing the camera on the desired measurement distance, a good calibration requires the detection of the calibration pattern from as close to as far as possible. This means that with a measuring field of 1 meter, not only from 0 to 1 meter, but from 0 to as far as possible must be covered by the calibration board.

Tip 2:

For a valid calibration frame it is important that the calibration pattern is always detected in the field of view of both cameras. Otherwise the frame is not counted.

Tip 3:

Calibration succeeds if all image corners are well covered by taking the calibration frames. So please take care that you are moving the calibration board as much around as possible.

Tip 4:

The more pictures are taken, the more precisely the camera will be calibrated. Should you be unhappy with the quality of any of the frames taken, you can delete them and add another individual captured frame. We recommend the capturing of 20 or more frames.

Tip 5:

Be patient because it is extremely important to work conscientiously and cleanly during calibration.

Now available: pattern projector for bin picking applications and challenging surfaces

With the pattern projector EFFI-LASE-V2 Nerian Vision expands its portfolio of stereo vision accessories and enables best 3D depth perception even in demanding machine vision applications.

Bin picking applications are currently the most requested applications in the field of automation technology. Which is not a surprising fact, because a solidly programmed solution saves time and money and will amortize its implementation in a very short time.

A bin picking solution is only considered attractive as long as it is affordable, easy to use, precise and fast. Additionally there are some main challenges such as occlusion, bad lighting conditions, reflections or smooth and textureless surfaces, which have to be taken into consideration while planning a bin picking application.

You can now improve the performance of your applications by combining our passive stereo vision image processing system with a pattern projector!

The pattern projector is delivered in an additional package and is not mounted on our 3D stereo vision system itself or on the stereo camera. This allows you to decide how and where the projector should be mounted. For example, in such a way that it does not blind the 3D stereo camera during operation when reflective surfaces need to be measured.

The projector can be used at different angles as long as it illuminates the desired field of view well. In addition, thanks to several stereo camera models and baseline distances, different measuring distances are conceivable – because our systems adapt to your application, not the other way around.

Another advantage is that different wavelengths of light can be emitted to get the best results from your application.

The following pictures were deliberately taken by us under conditions that are normally suboptimal for stereo vision (poor lighting, poor texture) in order to emphasize the performance improvement of the projector:

You can find the data sheet for the pattern projector here.

Convince yourself of the possibilities of 3D stereo vision using a pattern projector. Get your quote today:

Please send me a quote

Python support for SceneScan and new firmware version

Today our official Python API has been released together with a new major version of our SceneScan’s firmware.

Today there have been two important updates for our SceneScan system: a new major release of both, the firmware and the client software package. The most important change in the client software is that we now provide an official API for Python.

The Python programming language is becoming more and more popular in the machine vision domain, and we have had many requests for an official Python API. We are glad that today this new API is finally here, and available for both, Windows and Linux. The API comes with several examples and a demo application that show how to communicate with SceneScan and receive real-time 3D depth data. To get started, you can already browse the API documentation on our support website.

The other big update is firmware version 6.0 for SceneScan. This new version greatly simplifies system configuration. There is now a large set of pre-programmed presets for all common configurations with our Karmin2 and Karmin3 cameras. Hence, configuring SceneScan for one of these cameras can now be done with just a single click. Also, we significantly restructured the user interface. The less commonly used features are now grouped on the Advanced Settings section and the most important settings have now been moved to the new Acquisition Settings page.

We hope that these two releases will make it even easier to integrate SceneScan into your machine vision applications!