Efficient Global Optimization of Non-differentiable, Symmetric Objectives for Multi Camera Placement

TL;DR

This paper introduces a novel iterative method for globally optimizing camera placement and orientation in 3D scenes, effectively handling complex, non-differentiable objectives to enhance applications like reconstruction and surveillance.

Contribution

The paper presents a new globally convergent algorithm that leverages symmetry and surrogate functions for efficient multi-camera placement optimization.

Findings

Algorithm handles non-differentiable, expensive objectives.

Method accelerates optimization by exploiting symmetry.

Parallelizable and converges globally.

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.