Enhancing the analysis of video time series by means of a multi-agent architecture

1Citations
Citations of this article
2Readers
Mendeley users who have this article in their library.
Get full text

Abstract

In this paper we present a software architecture based on a multi-agent system whose major goal is the identification of traffic events from videos. In order to achieve this, H264/AVC motion vectors that appear in compressed video signal are taken as input. They are classified depending on their position in the scene and after that each group of motion vectors obtained from such classification is processed independently using statistical techniques. The use of this kind of techniques have been broadly used in the processing of time series like the one we take as input. After the statistical processing, individual results are compared between them in order to detect patterns related to possible traffic events. This comparison process can be understood as a cooperative process. So, to integrate the different processing components of this architecture we propose the use of a multi-agent system. Multi-agent systems allows to define a cooperative architecture using individual agents that can be run in parallel allowing to raise the performance and efficiency of the global process of event identification. The experimentation of this paper is driven to the detection of objects in complex traffic scenarios where the videos are captured from on-board cameras.

Cite

CITATION STYLE

APA

Rodriguez-Benitez, L., Giralt, J., Merino, D., Jimenez-Linares, L., & Moreno-Garcia, J. (2019). Enhancing the analysis of video time series by means of a multi-agent architecture. In Studies in Computational Intelligence (Vol. 796, pp. 11–21). Springer Verlag. https://doi.org/10.1007/978-3-030-00485-9_2

Register to see more suggestions

Mendeley helps you to discover research relevant for your work.

Already have an account?

Save time finding and organizing research with Mendeley

Sign up for free