Automatic Multi-Camera Setup Optimization for Optical Tracking

Abstract

Figure 1: Camera setup optimization sequence for a five sided CAVE with four cameras. The camera positions were constrained to remain on the open top side of the CAVE. Notice the increased point sample density in head height to improve the head-tracking robustness. ABSTRACTWe propose a method to determine the optimal camera alignment for a tracking system with multiple cameras by specifying the volume to be tracked and an initial camera setup. We use optimization strategies based on methods usually employed for solving nonlinear systems of equations. All approaches are fully automatic and take advantage of modern graphics hardware since we also implement a GPU-based, accelerated visibility test. The algorithm automatically optimizes the whole setup by adjusting the given set of camera parameters. We can steer the optimization towards different goals depending on the desired application, e.g. the widest possible volume coverage or maximum camera visibility to overcome heavy occlusion problems during the tracking process. We also consider parameter constraints that the user may specify according to restrictions in the local environment where the cameras have to be mounted. This allows for a convenient definition of higher level constraints for the camera setup.

Optimal stereo camera placement under spatially varying resolution requirements

“…Cowan et al [2] have experimented with methods to place multiple sensors and overcome the occlusion problems associated with 3D objects. Recently, Cerfontaine et al [1] have proposed a method to determine the optimal camera alignment for a tracking system with multiple cameras by specifying the volume to be tracked and an initial camera setup. Although these works have introduced the concept of high priority 3D volumes, most of these camera placements are for person tracking applications with 2D cameras rather than 3D stereo cameras and hence are focusing more on occlusion due to field of view than on accuracy and density of reconstructed 3D information.…”

Predicting and estimating the accuracy of n-occular optical tracking systems

“…In particular, we assume that a feature point can always be detected uniquely. We do not consider that some marker targets would not be uniquely detectable due to symmetries, visibility [12,20], or other properties the specific tracking algorithms relies on. The detection rate can be increased by better design -we predict only the accuracy once a feature is correctly identified.…”

“…We will show in section 4.2 how we have estimated this covariance. For real world setups like the one presented here, we additionally consider the field of view of the cameras and use only the cameras that are able to see the point for the error estimation [20].…”