7.5 Logging system
is assumed to be the function’s minimum. This threshold is set based on experimentation.
Extension to multiple cameras
The description of the algorithm thus far relates to one camera. With more than one camera there will be multiple intersection points when the camera axes are aligned correctly. Each point corresponds to the position of a camera. With multiple cameras, F (q) can be extended to
F (q) = C c=1 X 1 N c N X n =1
This calculates a combined variance of the intersection points for each camera. In this
equation c represents
represents the index of a camera, C represents the number of cameras, pn c
intersection point using the cth
is the mean of all of
the intersection points process can be used to
from the cth camera. Using this extension find a value of q that minimises F (q) and
the same minimisation each camera’s position
can be found in world coordinates by calculating
The logging system takes the output of the pose estimation algorithm and logs this to disk. This enables the data to be analysed in programs such as M TL B. The intermediate centroid data can also be logged to disk. Logging the centroid data to disk allows the pose to be estimated at a later date.
The user interface is a flexible environment that allows the user to visualise the centroids received from the camera modules and displays the pose information. The user inter- face uses a docking framework that allows windows to be displayed, moved, docked, and closed.
This section briefly describes the implementation of the Soft Hub in C#.
The software is
based around the Model View Controller pattern .
This is a widely accepted pattern
that allows a user interface to be decoupled from the underlying business logic.
name implies the pattern has three parts. These are the model, the views, and their con- trollers. The model encompasses the core functionality of the application and is composed
of problem domain classes and support classes such as the logging system.