A novel approach to manage the complexity and heterogeneity of video surveillance systems

Abstract

A novel approach to manage the high variability level of Video Surveillance Systems induced by emerging complexity and heterogeneity without compromise the system performance is introduced in this paper. Instead of using a modular architecture based on filters that use dynamic programming techniques, such as, inheritance, virtual functions and plug-ins, static programming techniques like template metaprogramming were used. Firstly, feature models are employed to represent the common and variable features at specification level. Finally, template metaprogramming is used to manage variability at implementation level. In opposition to modular architectures, this approach achieves a very high flexibility of management of hot spots and a high level of system performance trade off. Furthermore, in this solution, only the required code for a specific system configuration is compiled. This is the perfect technique for systems with serious memory constraints as embedded systems.