NOVATEUR PUBLICATIONS

International Journal of Research Publications in Engineering and Technology [IJRPET]

ISSN: 2454-7875
VOLUME 3, ISSUE 5,May -2017
SURVEY ON MOVING OBJECT TRACKING AND DETECTION
DHANASHRI V. MADHEKAR
M.E.(VLSI and Embedded System) G.H.R.I.E.T. Wagholi, Pune

PROF. MRINAL R. BACHUTE

M.E.(VLSI and Embedded System) G.H.R.I.E.T. Wagholi, Pune

ABSTRACT: mechanisms [1].Image Processing Hardware will acquire

Object tracking is an important component of images captured by camera after some pre defined
many computer vision systems. It is widely used in interval of time. Then it will processes much captured
number of fields such as video surveillance, robotics, image for detecting intrusion. If intrusion is detected
medical imaging, and human computer interface. To Image Processing Hardware will extract the features of
identify and tracking the real time object is important that intruding object and compare them with features of
concept in computer vision. In the paper, image objects stored in database. We have collected database for
processing algorithms are used for tracking a moving the objects those are to be destroyed. If match between
video object. The image processing algorithms used intruding object and one of the objects from database is
are (a)Noisy video generation with real time object found object is said to be recognized. System will track
motion (b)Video image back ground subtraction that object to calculate its velocity of motion. This velocity
(c)Video image these holding (d)Video image region information is needed to decide the angle and time in stant
filling(e)Video image height and width calculation (f) at which projectile is to be launched at intruding object to
Video image center computation (g) Video image and destroy it. Position of the intruding object in the form of x-
center image over lay. The image processing yco-ordinate is found and sent to microcontroller.
algorithms are developed initially by Model Based Microcontroller will control the angle of rotation of two
Design Approach using Simulink models of Matlab Servo Motors to position the can non aiming at the
Tool. Then these algorithms can be implemented on intruding object. At last cannon will get fired.
Raspberry Pi device. Object tracking is an important task within the field of
KEYWORDS: Raspberry pi, Simulink, Video image computer vision. The proliferation of high- powered
processing, tracking. computers, the availability of high quality and in expensive
video cameras, and the increasing need for automated
1. INTRODUCTION: video analysis has generated a great deal of interest in
The system is proposed for the purpose of intrusion object tracking algorithms. In its simplest form, tracking
detection, track and destroy the intruding object. The can be defined as the problem of estimating the trajectory
system will be mounted at some suitable place from which of an object in the image plane sit moves around a scene.
complete and clear view of the area under surveillance can In other words, a tracker assigns consistent labels to the
be captured with camera. Thus the image will be captured; tracked objects in different frame sofa video [2].
processed and desired action will be performed on it. Additionally, depending on the tracking domain, a
Object segmentation separates regions of interest in image tracker can also provide object centric information, such
data that identify real world objects. Segmenting and as orientation, area, or shape of an object.
tracking regions of arbitrary size within a scene allow the For Object Tracking and Classification basic thing is to
application to focus on more complex tasks like object obtaining an initial ask for a moving object, and to
recognition within a smaller spatial domain of the entire preprocess the mask. Normally the mask is affected by
spatial scene which reduces the processing time required “salt –and -pepper” noises. We apply morphological filters
to identify the object of interest. Reducing the spatial based on combinations of dilation and erosion to reduce
domain of the image decreases the computational the influence of noise, followed by a connected component
resources necessary for the detailed analyses required for analysis for labeling each moving object region.
object recognition. Very small regions are discarded. At this stage we
The system is provided with a higher solution camera, calculate the following features for each moving object
image processing hard ware, microcontroller, two region: bounding rectangle: the smallest aesthetic
servomotors and other supplementary hardware and rectangle that contains the object region. Kept record of
6|P a g e
 NOVATEUR PUBLICATIONS
International Journal of Research Publications in Engineering and Technology [IJRPET]
ISSN: 2454-7875
VOLUME 3, ISSUE 5,May -2017
the coordinate of the upper left position and the lower details the hardware implementation of real time motion
right position, what also provides size information (width segmentation algorithms on a FPGA Including the capture,
and height of each rectangle).Color: the mean R G B values processing and display stages. The proposed platform
of the moving object.Center: we use the center of the offers the flexibility of changing the processing modules,
bounding box as a simple approximation of the centroid of allowing the implementation of different motion
a moving object region.Velocity: defined as movement of segmentation algorithms and the online observation of the
number of pixels/second in both horizontal and vertical processing results. The system uses a minimum resource
direction. of cells allowing its implementation on low cost FPGAs.
In order to track moving objects accurately, especially
when objects are partially occluded, and the position of the B. Background Subtraction Algorithm for Moving
camera is not restricted to any predefined viewing angle, Object Detection in FPGA by J. Y. Mori [2]
these features are actually insufficient. We have to add Currently, both the market and the academic
further features that are robust and which can also be communities have required applications based on image
extracted even if partial occlusion occurs. and video processing with several real-time constraints.
One can simplify tracking by imposing constraints on On the other hand, detection of moving objects is a very
the motion and/or appearance of objects. For example, important task in mobile robotics and surveillance
almost all tracking algorithms assume that the object applications. In order to achieve an alternative design that
motion is smooth with no abrupt changes. One can further allows for rapid development of real time motion detection
constrain the object motion to be of constant velocity or systems, the paper proposes hardware architecture for
constant acceleration based on a priori information. Prior motion detection based on the background subtraction
knowledge about the number and the size of objects, or the algorithm, which is implemented on FPGAs (Field
object appearance and shape, can also be used to simplify Programmable Gate Arrays). For achieving this, the
the problem. following steps are executed: (a) a background image (in
gray-level format) is stored in an external SRAM memory,
GOALS AND OBJECTIVES: (b) a low-pass filter is applied to both the stored and
To design a system to detect intrusions, track current images, (c) a subtraction operation between both
intruding object and destroying it which will give complete images is obtained, and (d) a morphological filter is
and clear view of the area under surveillance and can be applied over the resulting image.
captured with camera. The system should be provided with Afterward, the gravity center of the object is calculated
a high resolution camera, image processing hardware, and sent to a PC (via RS-232 interface). Both the practical
microcontroller, two servo motors and other results of the motion detection system and synthesis
supplementary hardware and mechanisms. Camera should results have demonstrated the feasibility of FPGAs for
be placed at the fix point from where the area will be implementing the proposed algorithms on an FPGA based
covered clearly. hardware platform. The implemented system provides one
processed pixel per FPGA’s clock cycle (after the latency
MOTIVATION OF THE PROJECT: time) and speed-ups the software implementation (using
Motivation of the project comes from the loss of the real-time xPC Target OS from Math Works) by a factor
human being or soldiers during the war keeping of 32.
surveillance. Instead of humans we could replace our
system with camera and gun. C. FPGA Based Object Tracking System by Suhas
Jadhav[3]
2. LITERATURE SURVEY: The Paper presented an object tracking system based
A. An FPGA based Embedded Vision System for Real- on FPGA using canny edge detection Algorithm. The
Time Motion Segmentation by Jorge [1] system consists of canny edge detection algorithm
The proposed approach of an FPGA based real time implemented on FPGA kit to identify edges of real time
video processing platform for motion detection. The main object. Also a tracking and reorganization of object is done
advantage of FPGA based systems respect to processor by Smartphone camera. Canny edge detection algorithm is
based systems is its high performance when processing key stage in image processing and object recognizing
large amount of data flow, as video streams. The work application.

7|P a g e
 NOVATEUR PUBLICATIONS
International Journal of Research Publications in Engineering and Technology [IJRPET]
ISSN: 2454-7875
VOLUME 3, ISSUE 5,May -2017
The field programmable gate array contains logic time using the algorithms on the FPGA and the resulted
components that can be programmed to perform complex tracked video image output is displayed on a VGA monitor.
mathematical functions making them highly suitable for F. In 2003, Object Classification and Tracking in
the implementation of matrix algorithm. Individual frames Video Surveillance by Qi Zang and ReinhardKletteg
acquired from the target video are fed into the FPGA. [6]
These are subject to segmentation, thresholding, filtering In the paper, they have reviewed previous research on
stages. Following the object is tracked by comparing the moving object tracking techniques, analyze some
background frame and the processed updated frame experimental results, and finally provide our conclusions
containing the new location of the target. The result of the for improved performances of traffic surveillance systems.
FPGA implementation in tracking a moving object were One stationary camera has been used. Many applications
found to be correct and suitable for object tracking. have been developed for monitoring public areas such as
offices, shopping malls or traffic highways. In order to
D. Moving Object Detection and Tracking Based on control normal activities in these areas, tracking of
ZYNQ FPGA and ARM SOC by Wenchao Liu [4] pedestrians and vehicles play the key role in video
In the paper, FPGA-based moving object detection and surveillance systems. We classify these tracking
tracking system is introduced for image processing techniques into four categories:
application. The algorithms presented include object • Tracking based on a moving object region
detection based on dynamic background difference, • Tracking based on an active contour of a
kalman filter for object tracking. The target device for the moving object
implementation is a Xilinx Zynq-7000 FPGA. • Tracking based on a moving object model
The use of ARM dual core processor, NEON vector • Tracking based on selected features of
coprocessor unit and co-processing IP peripherals moving objects
generated by vivado HLS aim to improve the data Besides these four main categories, there are also
processing ability of the system. In the dynamic some other approaches on object tracking presents a
background, the system realized accurate detection and tracking method based on wavelet analysis.
tracking of moving objects in real-time.
G. In 2004, Moving Object Detection, Tracking and
E. Moving Object Tracking Application: FPGAand Classification for Smart Video Surveillance By Yi
Model Based Implementation Using Image githan Dedeoglu[7]
Processing Algorithms by Sofia Nayak [5] The paper,represents a smart visual surveillance
With increased resource size, powerful DSP blocks and system with real-time moving object detection,
large on-chip memory, Field Programmable Gate Array classification and tracking capabilities. The system
(FPGA) devices play a major role as hardware platforms operates on both color and gray scale video imagery from
for implementing compute intensive video image a stationary camera. In the proposed system moving object
Processing applications. In this paper, image processing detection is handled by the use of an adaptive background
algorithms are used for tracking a moving video object. subtraction scheme which reliably works in indoor and
The image processing algorithms used are (a) Noisy video outdoor environments. We also present two other object
generation with random motion (b) Video image median detection schemes, temporal differencing and adaptive
filter (c) Video image back ground removal (d) Video background mixture models, for performance and
image thresholding (e) Video image edge detection (f) detection quality comparison. The proposed system is able
Video image height and width calculation (g) Video image to distinguish transitory and stopped foreground removed
center computation (h) Video image and center image objects; classify detected objects into different groups such
overlay. The image processing algorithms are developed as human, human group and vehicle; track objects and
initially by Model Based Design Approach using Simulink generate trajectory information even in multi-occlusion
models of MATHWORK’s MATLAB Tool. Then these cases and detect fire in video imagery.
algorithms are implemented on ALTERA CYCLONE-II FPGA System is assumed to work real time as a part of
device using TERASIC DE2 FPGA hardware kit and avideo-based surveillance system. The computational
ALTERA QUARTUS-II software tool. The input video image complexity and even the constant factors of the algorithms
is taken From a NTSC/PAL camera and processed in real we use are important for real time performance. Hence,

8|P a g e
 NOVATEUR PUBLICATIONS
International Journal of Research Publications in Engineering and Technology [IJRPET]
ISSN: 2454-7875
VOLUME 3, ISSUE 5,May -2017
our decisions on selecting the computer vision algorithms research works are taking place in this field.Availability,
for various problems are affected by their computational efficiency of usage along with the increasing popularity of
run time performance as well as quality. Furthermore, our video on internet and versatility of video application
system’s use is limited only to stationary cameras and heavily rely on object detection and tracking in videos.
video inputs from Pan/Tilt/Zoom cameras where the view Object detection provides a good range of accuracy at
frustum may change arbitrarily are not supported. The different scenarios based on their application.
system is initialized by feeding video imagery from a static In the paper, an overview of recent research activities
camera monitoring a site. Most of the methods are able to in a video tracking systems has been presented. In video
work on both color and monochrome video imagery. tracking systems, the researchers will cope with the
problems of visual content extraction and image
3. PROPOSED WORK: understanding that several of techniques have been
proposed in detection, tracking, object representation and
recognize human behaviors.
Furthermore, for the future of video tracking systems,
enlarge and enhance the view of a tracking system with
multiple cameras should be considered. In this issue we
should deal with environment situation in different object
views, a huge volume of video data recorder and real-time
processing.

5. ACKNOWLEDGEMENT:
Research paper Author is heartily thankful to Guide “Prof.
Mrs. Mrinal R. Bachute”, G. H. Raisoni Institute of
Engineering and Technology, Pune for guidance,
inspiration and her valuable time.
Figure 1: Flow Diagram of Proposed Work I am also extremely thankful to Principal “Prof. Dr. R.D.
Kharadkar” and HOD of electronics and
Raspberry-pi sends command to camera to capture telecommunication department and also thankful to P.G.
image. Image Processing Algorithm running in raspberry- Co- ordinator “Prof. Mrs. M. R. Bachute” for their valuable
pi will acquire images captured by camera after some guidance.
predefined interval of time. Then it will process every
captured image for detecting intrusion. If intrusion is REFERENCES:
detected Image Processing Hardware will extract the 1) Jorge Hiraiwa, Enrique Vargas, Sergio Toral’’ An FPGA
features of that intruding object and compare them with based Embedded Vision System for Real-Time Motion
features of objects stored in database. We have collected Segmentation’’- 17th International Conference on
database for the objects those are to be destroyed. If match Systems, Signals and Image Processing in IWSSIP
is found between intruding object and the object from 2010.
database, object is said to be recognized. System will track 2) C. S´anchez- Ferreira, J. Y. Mori, C. H. Llanos
that object to calculate its velocity of motion. This velocity ‘’Background Subtraction Algorithm for Moving Object
information is needed to decide the angle and time instant Detection in FPGA’’ IEEE 2012.
at which projectile is to be launched at intruding object to 3) Suhas Jadhav, Rohit Narvekar, Ajay Mandawale,
destroy it. Position of the intruding object in the form of x- Sachin Elgandelwar’’ FPGA Based Object Tracking
y co-ordinate is found and alarm will be given System’’ Fifth International Conference on
Communication Systems and Network Technologies
4. CONCLUSION: in 2015.
The survey on object identification, detection and 4) Wenchow Liu, He Chen, Long Ma’’ Moving object
tracking techniques that is able to detect an object. Many detection and tracking based On zynqfpga and arm
research issues have been highlighted here. There are a lot soc’’.
of challenges in object detection and identification; still

9|P a g e
 NOVATEUR PUBLICATIONS
International Journal of Research Publications in Engineering and Technology [IJRPET]
ISSN: 2454-7875
VOLUME 3, ISSUE 5,May -2017
5) Sofia Nayak, Shashank Sekhar Pujari, ‘’Moving Object
Tracking Application: Fpga And Model Based
Implementation Using Image Processing Algorithms’’
International Conference on Computing
Communication Control and Automation,
ICCUBEA.2015.
6) Qi Zang and Reinhard Klette ’’Object Classification
and Trackingin Video Surveillance’’ N. Petkov and M.
A. Westenberg (Eds.): CAIP 2003, LNCS 2756, pp.
198–205, 2003.© Springer-Verlag Berlin
Heidelberg 2003.
7) Yigithan Dedeoglu ’’Moving Object Detection,
Tracking and Classification For Smart Video
Surveillance’’ August, 2004.
8) VM Sandeep Rao, Aravinnd Natarajan, S. Moorthi and
M.P.Selvan “Real-Time Object Tracking in a video
Stream using Field Programmable Gate Array”,
IEEE,2012.
9) Manoj Pandey, DorothiBorgohain, Gargi Baruah, J.S.
Ubhiand Kota Solomon Raju “Real Time Object
Tracking: Simulation and Implementation on FPGA
Based Soft Processor”, ICSSITE, 2013.

10 | P a g e