Optimal camera placement for an active-vision system

Image-analysis in an industrial environment can be simplified by premarking the objects using artificial markers and active vision to acquire prespecified views of these objects. In this context, a problem addressed is the determination of the initial locations of the cameras (including one mobile camera) such that marker(s) can be efficiently detected. A method for determining the optimal number of cameras and their best locations is presented, given the objects' geometries and the positions of markers on them. Such an optimal camera arrangement would yield maximum detectability of markers and also satisfy a user given threshold on the probability of detecting at least one marker. Analytic-geometry, probability theory and CAD databases are used in solving this problem.<<ETX>>

[1]  A. N. Babii.  Algorithm for finding the value of the global extremum of a function of several variables with a given accuracy , 1978, Cybernetics.

[2]  Hugh F. Durrant-Whyte,et al.  A Bayesian Approach to Optimal Sensor Placement , 1990, Int. J. Robotics Res..

[3]  Beno Benhabib,et al.  Pre-marking methods for 3D object recognition , 1989, Conference Proceedings., IEEE International Conference on Systems, Man and Cybernetics.

[4]  Roger Y. Tsai,et al.  Model-Based Planning of Sensor Placement and Optical Settings , 1990, Other Conferences.

[5]  Gregory D. Hager,et al.  Task-Directed Sensor Fusion and Planning , 1990 .

[6]  Robert M. Haralick,et al.  Automatic sensor and light source positioning for machine vision , 1990, [1990] Proceedings. 10th International Conference on Pattern Recognition.

[7]  Gregory D. Hager.  Choosing Viewpoints and Features , 1990 .