marker

Michael Krause

Source code is available in our GitHub repository: https://github.com/InstituteOfErgonomics/marker

No installation needed just download/copy all dll/exe files from folder ‘Release’. GitHub also provides a complete zip archive.

The open source tool provided here could help in some situations, when working with the older Dikablis/D-Lab 2 versions. The official tool chain has a marker detector, that tries to detect landmarks (two dimensional bar codes). These landmarks are used in analysis to define areas of interest. The tool presented here, can be seen as an alternative marker detector without predefined landmarks. The original marker detector needs the white frame around the codes clearly visible without much variance in white hue (shadows), also sometimes challenged by overexposure and blurry images, e.g. when the head moves (even in exhaustive mode).

Despite diligent previous testing, in one driving simulator experiment 24 data sets were rendered useless due to different nuisance. That was the involuntary motivation to code the tool.

The tool let you select a distinctive part of the video scene as template and uses simple template matching to find the location of the template in every video frame. For the intented experiment the illuminated speedometer (LCD) was an excellent template, with its unique and contrasty shapes (speed & RPM round scales).

The (smallest difference) template matching has some drawbacks that you need to know before:

1) It uses always the best fit location for the template in every video frame, even if the template is not visible in the frame. You have to adjust the threshold value appropiate. The threshold is a normalized value between 0-1000 and 0 means a perfect fit.

2) The template matching is not scale and rotation invariant. But to some extend robust against small angles and size changes.The original marker detector detects the four corners of the 2D codes. These corners are later used to rotate and scale areas of interest by the original tools. The tool here, simply draw a constant small rectangle (20×20 pixel) in the left upper corner were it finds the best fitting location for the template.

3) The results can be saved to a marker file. Therefore you have to specify the cryptic name of one marker (e.g. ’0F31′). Preferably one that is not used in your project. If the file is already present, the tool choose an own name. So carefully read the message when clicking the save button and maybe rename the file afterwards. If your project is already imported into D-Lab in some situations you have to import again.

Use at your own risk. Make backups of your experimental data before.

<iframe width="560" height="315" src="https://www.youtube.com/embed/bTqgrwlgl_c" frameborder="0" allowfullscreen></iframe>

Dikablis and D-Lab are trademarks/products (eye tracking) of Ergoneers GmbH http://www.ergoneers.com They are used in the documents for description (nominative use). This is NOT an official Ergoneers tool!

This documentation is not in the public domain, but the following is valid for it:

THE DOCUMENTATION IS PROVIDED “AS IS”, WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE DOCUMENTATION OR THE USE OR OTHER DEALINGS IN THE DOCUMENTATION.