Hybrid Attributes Technique Filter for the Tracking of Crowd Behavior

Hybrid Attributes Technique Filter for the Tracking of Crowd Behavior

Hocine Chebi
Copyright: © 2021 |Pages: 10
DOI: 10.4018/978-1-7998-6659-6.ch003
OnDemand:
(Individual Chapters)
Available
$33.75
List Price: $37.50
Current Special Offers
10% Discount:-$3.75
TOTAL SAVINGS: $3.75

Abstract

In this chapter, the authors propose two algorithms based on the device of attributes for tracking of the abnormal behavior of crowd in the visual systems of surveillance. Previous works were realized in the case of detection of behavior, which uses the analysis and the classification of behavior of crowds; this work explores the continuity in the same domain, but in the case of the automatic tracking based on the techniques of filtering one using the KALMAN filter and particles filter. The proposed algorithms he the technique of filter with particle is independent from the detection and from the segmentation human, so is strong with regard to (compared with) the filter of Kalman. In conclusion, the chapter applies the method for tracking of the abnormal behavior to several videos and shows the promising results.
Chapter Preview
Top

State Of The Art Of Tracking

Tracking methods propose to recognize and locate over time objects present in a temporal sequence of images (Gabriel 2003). In the context of crowds, they find a particular interest in video surveillance where the tracking of individuals makes it possible to automatically control the comings and goings in a space. Like image recognition, tracking can rely on graphic properties such as colors or outlines (Mathes 2006; Yang 2005). The added temporal dimension also makes it possible to assume continuity in the presence and position of people in the scene, despite the occlusions (Rabaud 2006).

In this part, we will therefore explore four main approaches: follow-up by regional approach, follow-up using a model, follow-up by contours approach, follow-up using attributes. For each of these methods, it must be kept in mind that a good monitoring method must be robust, precise and above all fast to be able to follow a person in real time. The quality of monitoring is also dependent on good detection of people on the move. We will present the different monitoring techniques, in order to draw the different advantages and disadvantages:

Complete Chapter List

Search this Book:
Reset