Motion and Feature Based Person Tracking
In Surveillance Videos


  Surveillance Videos.pdf (Size: 1.12 MB / Downloads: 11)

Abstract

This work describes a method for accurately
tracking persons in indoor surveillance video stream obtained
from a static camera with difficult scene properties including
illumination changes and solves the major occlusion problem.
First, moving objects are precisely extracted by determining
its motion, for further processing.

INTRODUCTION

Moving Objects Detection and tracking are widely
used low-level tasks in many computer vision applications,
like surveillance, monitoring, robot technology, gesture
recognition, object recognition etc. Many approaches have
been proposed for moving object detection and tracking
from videos, mainly dedicated to traffic monitoring and
visual surveillance.

RELEVANT WORK

We survey the techniques and method relevant to
object tracking, specifically approaches that perform
feature based tracking and handle occlusions. For accurate
tracking, the motion must be accurately detected using
suitable methods, but they are affected by a number of
practical problems such as shadow and lighting change
over time.
Many researchers have given their contributions to
Motion based object detection and tracking under both
indoor and outdoor scenes and provide solutions to the
above mentioned problems.

PROPOSED METHODOLOGY

Our algorithm aims to assign consistent identifier to
each object appears in scene when individual merge into or
split from the group and involves several methods to obtain
the lowest possibility of false tracking and tagging. In
tracking interested object (human), shadows affect the
performance of tracking and leads to false tagging. To
avoid this problem, we apply mean filter to remove noise
which causes the image sequence to blur. Since we are
using color information for tracking, blurring causes no
loss of data. The structural design of our proposed method
shown in Fig.1

CONCLUSION

The advantages of using color as feature to
achieve object’s similarity is analyzed and found that it is
robust against the complex, deformed and changeable
shape (i.e. different human profiles). In addition, it is also
scale and rotation invariant, as well as faster in terms of
processing time. Color information is extracted, stored and
compared to find uniqueness of each object.