Real Time Motion Detection Video Surveillance For Categorizing Between Human and Non-Human

Abstract

An Automated Video Surveillance and
monitoring system has a rich history. Traditional video
surveillance systems are large in number but still
retrenched in an extensive manner. The system targets
at tracking an object and segregates it as Human or
Non-Human entities, wherein the non-human entities
would be further analysed into its respective categories
which would help in subsequent analytics.The system
when recognizes suspicious activity, it is captured
instantly and the alarm is triggered for the security
purpose. It is necessary to introduce an application that
automatically formulates the images captured in order
to detect precarious situations or undesirable
encroached objects. Object detection is a mandatory
step in automated video surveillance. Foreground
extraction in harmony with background subtraction is
further collaborated with the thresholded image for
revelation of entities. The system engages a
contemporary combination of Background Modelling,
Support vector machine (SVM) and a Human Detection
for Surveillance (HDS) System. The HDS system
assimilates a Histogram of Oriented Gradients based on
a human detector which is in limelight for its
performance in detecting humanistic appearances.
Detailed analysis is carried out on the performance of
the system on various test videos.