Metaheuristics-Assisted Placement of Omnidirectional Image Sensors for Visually Obstructed Environments.

Area of Science:

• Computer Vision
• Artificial Intelligence
• Robotics

Background:

• Optimal camera placement (OCP) is vital for effective surveillance.
• Existing solutions often lack comprehensive coverage maximization, especially for omnidirectional cameras in complex indoor environments.

Purpose of the Study:

• To investigate the efficacy of various metaheuristic optimization algorithms (MOAs) for OCP of omnidirectional cameras.
• To identify which MOAs perform best for maximizing surveillance coverage in indoor, partially occluded settings.

Main Methods:

• Evaluated popular MOAs including Particle Swarm Optimization (PSO), Grey Wolf Optimizer (GWO), and Genetic Algorithms (GA).
• Conducted experiments using two distinct indoor layouts with varying obstructions and two omnidirectional camera models.

Main Results:

• The performance of MOA-based OCP is highly dependent on the algorithm's specific search strategy.
• Certain MOAs demonstrated superior performance in maximizing surveillance coverage compared to others.

Conclusions:

• The choice of metaheuristic algorithm and its search mechanism is critical for successful omnidirectional camera placement.
• This research provides insights into selecting appropriate MOAs for enhanced indoor surveillance systems.