1. Efficient Global Optimization of Non-differentiable, Symmetric Objectives for Multi Camera Placement
- Author
-
Maria L. Hänel, Carola-Bibiane Schönlieb, Hanel, ML [0000-0001-7013-3544], and Apollo - University of Cambridge Repository
- Subjects
FOS: Computer and information sciences ,Mathematical optimization ,Iterative method ,Computer science ,G.1.6 ,Computer Vision and Pattern Recognition (cs.CV) ,Computer Science - Computer Vision and Pattern Recognition ,G.1.1 ,I.4.8 ,I.6 ,I.2.10 ,I.2.11 ,I.2.8 ,01 natural sciences ,FOS: Mathematics ,90-05, 90C26, 90C30, 90C56, 90C59 ,Computer Science - Multiagent Systems ,Differentiable function ,Electrical and Electronic Engineering ,Instrumentation ,Global optimization ,Mathematics - Optimization and Control ,40 Engineering ,Parallelizable manifold ,010401 analytical chemistry ,3D reconstruction ,Solver ,Stationary point ,0104 chemical sciences ,Cost reduction ,Optimization and Control (math.OC) ,Multiagent Systems (cs.MA) - Abstract
We propose a novel iterative method for optimally placing and orienting multiple cameras in a 3D scene. Sample applications include improving the accuracy of 3D reconstruction, maximizing the covered area for surveillance, or improving the coverage in multi-viewpoint pedestrian tracking. Our algorithm is based on a block-coordinate ascent combined with a surrogate function and an exclusion area technique. This allows to flexibly handle difficult objective functions that are often expensive and quantized or non-differentiable. The solver is globally convergent and easily parallelizable. We show how to accelerate the optimization by exploiting special properties of the objective function, such as symmetry. Additionally, we discuss the trade-off between non-optimal stationary points and the cost reduction when optimizing the viewpoints consecutively., Comment: Submitted to be reviewed, 10 pages, 6 figures, 2 tables, 3 algorithms
- Published
- 2021
- Full Text
- View/download PDF