Redundant Visual Coverage of Prioritized Targets in IoT Applications

• Elivelton O. Rangel UEFS
• Daniel G. Costa UEFS
• Angelo Loula UEFS

Resumo

Visual monitoring is an important service for various emerging and traditional applications in the Internet of Things (IoT) field, since still images and video streams can be gathered and processed for different kinds of multimedia-based tasks. Frequently, a set of targets may need to be covered by interconnected cameras, providing concurrent multiple views under different perspectives for the applications, besides enhanced resistance to camera failures. In this context, if targets have different monitoring priorities, the configuration of the cameras can be optimized, demanding proper algorithms to maximize redundant coverage over more relevant targets. This paper proposes a lightweight greedy algorithm and a costly but more effective evolutionary algorithm to optimize redundant visual coverage by cameras, both aimed at reduction of uncovered targets and maximization of redundant coverage over the most relevant targets, which may improve the overall quality of different visual IoT applications.

Palavras-chave: Visual IoT applications, Camera calibration, Prioritized target viewing, Greedy algorithms, Evolutionary algorithms