BLOB TRACKING MODEL USING VIDEO OBJECT

TRACKING IN THE COMPRESSED DOMAIN

S.Rajasekaran#1,
#1
II-M.E, Department of VLSI DESIGN, Sasuire college of engineering and
technology, vijayamangalam.
rajasekar0107@gmail.com
N.Kannapiran M.E., *2
Asst professor, Department of ECE, , Sasuire college of engineering and
technology, vijayamangalam.
ABSTRACT
Despite the recent progress in both
pixel-domain and compressed-domain
video object tracking, the need for a
tracking framework with both reasonable
accuracy and reasonable complexity still
exists. This paper presents a method for
tracking moving objects in H.264/AVCcompressed video sequences using a
spatio-temporal Markov random field
(STMRF) model. An ST-MRF model
naturally integrates the spatial and
temporal aspects of the objects motion.
Built upon such a model, the proposed
method works in the compressed domain
and uses only the motion vectors (MVs)
and block coding modes from the
compressed bit stream to perform
tracking. First, the MVs are pre processed
through intra coded block motion
approximation and global motion
compensation. At each frame, the decision
of whether a particular block belongs to
the object being tracked is made with the
help of the ST-MRF model, which is
updated from frame to frame in order to
follow the changes in the objects motion.
The proposed method is tested on a
number of standard sequences, and the
results demonstrate its advantages over
some of the recent state-of-the-art
methods.

INTRODUCTION
Object tracking can be defined as the
process of segmenting an object of interest
from a video scene and keeping track of its
motion, orientation, occlusion etc. in order
to extract useful information. Object
tracking in video processing follows the
segmentation step and is more or less
equivalent to the recognition step in the
image processing. Detection of moving
objects in video streams is the first relevant
step of information extraction in many
computer vision applications, including
traffic monitoring, automated remote video
surveillance, and people tracking. There are
basically three approaches in object
tracking. Feature-based methods aim at
extracting characteristics such as points, line
segments from image sequences, tracking
stage is then ensured by a matching
procedure at every time instant. Differential
methods are based on the optical flow
computation, i.e. on the apparent motion in
image sequences, under some regularization
assumptions. The third class uses the
correlations to measure inter image
displacements. Selection of a particular
approach largely depends on the domain of
the problem.

EXISTING SYSTEM ANALYSIS:

From the discussion, in the existing
paper it can be seen that object tracking has
many useful applications in the robotics and
computer vision fields. Several researchers
have explored and implemented different
approaches for tracking. The success of a
particular approach depends largely on the
problem domain.
In other words, a method that is
successful in robot navigation may not be
equally
successful
in
automated
surveillance. Further there exists a cost or
performance trade off. For real time
applications we may need a fast high
performance system on the other hand
offline applications we may use a relatively
cheap (and slower in performance). It can
also be seen from the diverse nature of the
techniques used that the field has a lot of
room for improvement.
Differential
methods are based on the optical flow
computation, i.e. on the apparent motion in
image sequences, under some regularization
assumptions. The third class uses the
correlation to measure inters image
displacements. Selection of a particular
approach largely depends on the domain of
Video tracking is the process of locating a
moving object (or multiple objects) over
time using a camera. It has a variety of uses,
some of which are human-computer
interaction, security and surveillance, video
communication
and
compression,
augmented reality, traffic control, medical
imaging and video editing. Video tracking
can be a time consuming process due to the
amount of data that is contained in video.
Adding further to the complexity is the
possible need to use object recognition
techniques for tracking. Object tracking in
video processing follows the segmentation
step and is more or less equivalent to the
recognition step in the image processing.
Detection of moving objects in video
streams is the first relevant step of

information extraction in many computer

vision applications.

Moving object segmentation in

compressed domain plays an important role
in many real-time applications, e.g. video
indexing, video transcending, video
surveillance, etc. Because H.264/AVC is the
up-to-date video-coding standard, few
literatures have been reported in the area of
video analysis on H.264/AVC compressed
video.In contrast to prior video coding
standards, DC coefficients in intra-coded
pictures no longer represent average energy.
But only represent an energy difference.
Algorithms working on DC coefficients can
therefore no longer be applied to
H.264/AVC bit streams. Furthermore
H.264/AVC supports variable block size
motion compensation. A macro block can
now be partitioned into several smaller
blocks, where each block has its own MVs.
This is very different from former MPEG
video standards, where regular block size
MVs are used .we present a new object
detection algorithm working on H.264/AVC
which relies on MVs. In order to alleviate
the noisy motion vector field and, as a
consequence, to make the detection more
accurate, the reliability of the MVs is
estimated and incorporated during the

detection. A new algorithm for moving

object detection in the H.264/AVC
compressed domain. By estimating the
reliability of the MVs based on surrounding
frames noisy MVs are removed, while MVs
corresponding to true motion are preserved.
Future work includes optimal
selection of variable thresholds based on a
background model. Furthermore, to make
the detection results more accurate, DCT
coefficients will be considered as well.

previous work in fusing stereo with TOF

sensors, we focused on static scenes and
took several depth maps and compute the
mean. The result shows that it is less
susceptible to image noise. The mean depth
map was used as input. In Figure 6, we
visualize 3D points and depth maps using
one shot (low SNR) and the averaged of 10
shots (high SNR).

PROPOSED SYSTEM
In this proposed paper, we have
presented a novel approach to track a
moving object in a H.264/AVC-compressed
video. The only data from the compressed
stream used in the proposed method are the
motion vectors and block coding modes. As
a result, the proposed method has a fairly
low processing time, yet still provides high
accuracy. After the pre processing stage,
which consists of intra-coded block motion
approximation
and
global
motion
compensation, we employ Spatio-Temporal
Markov Random Field model to detect and
track a moving target. Using this model, an
estimate of the labeling of the current frame
is formed based on the previous frame
labelling and current motion information.
The results of experimental evaluations on
ground truth video demonstrate superior
functionality and accuracy of our approach
against other state of the art compresseddomain segmentation/tracking approaches.

We can clearly see the depth map

from multiple shots is better than that from
the single shot. However, simple average
over multiple shots can be applied only to
static scenes. Our STMRF method uses
temporal correspondences to reduce noises
and improve depth estimate in dynamic
scenes. In this way, it can also be viewed as
a temporal denoising scheme when applying
to the TOF sensor alone. We evaluate our
STMRF method on a number of real scenes.
For each scene, we compare depth maps and
their geometrical representations (3D point
clouds). All of them are using the left
camera as the reference view.

Although our algorithm works well

even with fixed parameter values, possibly
better performance may be obtained by
adaptive tuning, although this would in
general increase the complexity .Before
comparing results from SMRF and STMRF,
we did a simple test to show that the quality
of depth maps from SMRF method is
influenced by the Signal to Noise Ratio
(SNR) of the TOF sensor. In one of our

EVALUATION

To make a fair comparison, we set

the message truncation value Tm = 0:3, and
we set the TOF ssensor
capturing
frequency around 20 FPS and the stereo rig
around 100 FPS. All compared frames are
using the TOF sensors time stamp as the
reference. The system specification

includes the Software requirements for

the simulation of the project. The project

is simulated with the help of a Desktop

Computer or a Laptop. The development
tool used for the simulation of this
project is MATLAB 7.0

FUTURE ENHANCEMENT
In this paper, the 3D MRF-based
detector deals well with difficult motion
detection and tracking problems such as
objects with uniform color and objects
camouflaged by similar appearance to the
background. The approach is validated on
several synthesized and real-world video
sequences. So far, only motion information
is smoothed to estimate the current state. In
future work, other cues such as shape and
appearance can be fused into the tracking A
PARTICULAR moving objects in SPIHT
compressed video sequences using Blobs.

CONCLUSION

REFERENCES
[1] Arvanitidou, M. Tok, A. Krutz, and T.
Sikora, Short-term motionbased object
segmentation, in Proc. IEEE Int. Conf.
Multimedia Expo, Barcelona, Spain, Jul.
2011, pp. 16 .
[2]. Aeschliman, J. Park, and A. C. Kak, A
probabilistic
framework
for
joint
segmentation and tracking, in Proc. IEEE
Comput.
Vis.
PatternRecognit.,
San
Francisco, CA, Jun. 2010, pp. 13711378
[2].
[3]Astola, P. Haavisto, and Y. Neuvo,
Vector median filters, Proc. IEEE, vol.
78, no. 4, pp. 678689, Apr. 1990 [36].
[4].Comaniciu, V. Ramesh, and P. Meer,
Kernel-based object tracking,IEEE Trans.
Pattern Anal. Mach. Intell., vol. 25, no. 5,
pp. 564575,May 2003 [3].
[5].Besag, On the statistical analysis of
dirty pictures, J. Royal Stat.Soc. B, vol. 48,
no. 3, pp. 259302, 1986 [29].
[6].Cherian, J. Andersh, V. Morellas, N.
Papanikolopoulos,
and
B.Mettler,
Autonomous altitude estimation of a UAV
using a singleonboard camera, in Proc.
IEEE Int. Conf. Intell. Robots Syst.,
Oct.2009, pp. 39003905..
[7]. Chen and I. V. Bajic, A joint approach
to global motionestimation and motion
segmentation from a coarsely sampled

To detect moving objects within a

sequence of video frames, we construct a
MRF model such that each pixel
s hidden
state (motion likelihood) has a 6-connected
spatio-temporal neighborhood. a novel
approach to track a moving object in a
H.264/AVC-compressed video.
The only data from the compressed stream
used in the proposed method are the motion
vectors and block coding modes. As a result,
the proposed method has a fairly low
processing time, yet still provides high
accuracy.After the preprocessing stage,
which consists of intra-coded block motion
approximation
and
global
motion
compensation, we employ Spatio-Temporal
Markov Random Field model to detect and
track a moving target. Using this model, an
estimate of the labeling of the current frame
is formed based on the previous frame
labeling and current motion information.The
results of experimental evaluations on
ground truth video demonstrate superior
functionality and accuracy of our approach
against other state-of the- art compresseddomain segmentation/tracking approaches.

motionvector field, IEEE Trans. Circuits

Syst. Video Technol., vol. 21, no. 9,pp.
13161328, Sep. 2011.
[8].Chen, I. V. Bajic, and P. Saeedi,
Moving
region
segmentationfrom
compressed video using global motion
estimation and Markovrandom fields, IEEE
Trans. Multimedia, vol. 13, no. 3, pp. 421
431,Jun. 2011.
[9].Chen, I. V. Bajic, and P. Saeedi, Motion
segmentation incompressed video using
Markov random fields, in Proc. IEEE Int.
Conf.Multimedia Expo, Singapore, Jul.
2010, pp. 760765.
[10].Fei and S. Zhu, Mean shift clusteringbased moving object segmentationin the
H.264 compressed domain, IET Image
Process., vol. 4,no. 1, pp. 1118, Feb. 2010
[11].Geman and D. Geman, Stochastic
relaxation, Gibbs distributions, and the
Bayesian restoration of images, IEEE
Trans. Pattern Anal. Mach.Intell., vol. 6, no.
6, pp. 721741, Nov. 1984