Hostname: page-component-7b7567c8c-fvjsw Total loading time: 0 Render date: 2026-11-16T06:29:21.904Z Has data issue: false hasContentIssue false

Object Extraction and Classification in Video Surveillance Applications

Published online by Cambridge University Press:  19 December 2016

Muhsin Civelek
Affiliation:
Department of Computer Engineering, Turkish Military Academy, Ankara, Turkey. E-mail: mcivelek@kho.edu.tr
Adnan Yazici
Affiliation:
Department of Computer Engineering, Middle East Technical University, Ankara, Turkey. E-mail: yazici@ceng.metu.edu.tr
Rights & Permissions [Opens in a new window]

Abstract

In this paper we review a number of methods used in video surveillance applications in order to detect and classify threats. Moreover, the use of those methods in wireless surveillance networks contributes to decreasing the energy consumption of the devices because it reduces the amount of information transferred through the network. In this paper we focus on the most popular object extraction and classification methods that are used in both wired and wireless surveillance applications. We also develop an application for identification of objects from video data by implementing the selected methods and demonstrate the performance of these methods on pre-recorded videos using the outputs of this application.

Information

Type
In Honour of Erol Gelenbe
Copyright
© Academia Europaea 2016 
Figure 0

Figure 1BS operation.

Figure 1

Figure 2Application of morphological operations on foreground.

Figure 2

Figure 3Segmented foreground.

Figure 3

Table 1Summary of the features and classification methods used in this study.

Figure 4

Table 2Performance results

Figure 5

Figure 4Similarity matching using SURF key points.

Figure 6

Figure 5Popular classifiers.

Figure 7

Figure 6C++ application for object extraction and classification from surveillance videos.