Estimates a 2-D approximation of the volumetric scene reconstruction
error for all possible object locations in the specified area using
the specified camera location.
This can be applied to optimize the placement of the cameras in an
environment (with respect to the expected reconstruction error) as
well as to optimize the view directions of the cameras online (since
it is very efficient to calculate; a C/C++ implementation is provided).
Usage example:
Ae=estimate_error_2d_map([0 0 3.8 6.6],
[0.25 3.5 0.25 3.5; 0.25 0.45 3 4],
0.2)
Where the first argument specifies the modeled room area, the
second parameter contains the locations of the cameras in the
room and the third is the (hypothetical) radius of the target
object. Please refer to [1] for further details.
It has been applied in [1] and [2] for offline optimization of the
camera locations and to optimize the camera view directions online:
[1] B. Schauerte, J. Richarz, et al., "Multi-Modal and Multi-Camera
Attention in Smart Environments". In Proceedings of the 11th
International Conference on Multimodal Interfaces (ICMI), 2009.
[2] B. Schauerte, T. Plötz, G. A. Fink, "A Multi-modal Attention
System for Smart Environments". In Proceedings of the 7th
International Conference on Computer Vision Systems (ICVS), 2009. |
