Kq3518291832

Mayur Salve et al Int. Journal of Engineering Research and Application
ISSN : 2248-9622, Vol. 3, Issue 5, Sep-Oct 2013, pp.1829-1832

www.ijera.com

RESEARCH ARTICLE

OPEN ACCESS

Fpga Based Moving Object Detection Algorithm Implementation
for Traffic Surveillance
Mayur Salve, Mahesh Rajput, Sanket Shingate
(Department of Electronics, Mumbai University, India

ABSTRACT
The main objective of this paper is detection of vehicles by background subtraction technique and the automatic
surveillance of traffic based on the number of vehicles. The algorithm takes into consideration three main
techniques namely Background Subtraction, Edge Detection and Shadow Detection. Background Subtraction
block is sub-divided into Selective and Non-selective parts to improve the sensitivity and give accurate
background. Edge detection helps to detect the exact boundaries of moving vehicles. This is followed by the
shadow detection block that removes the falsely detected pixels that are generated due to shadow of the vehicle.
By analyzing the output of the blocks discussed above, the final mask is generated. The mask along with the
input frame is processed to give the final output frame where the detected object is highlighted. Furthermore,
parameters such as number of blobs per frame (vehicles) and the area of blobs can be used for traffic
surveillance. The algorithm of object detection is implemented on FPGA using VHDL. Spartan-6 Development
Board is used for implementation of the same.
Keywords – FPGA, Image Processing, SPARTAN6, Traffic surveillance, VHDL.

I.

INTRODUCTION

Various type of traffic surveillance systems
are often used for controlling traffic and detecting
unusual situations, such as traffic congestion or
accidents. This paper describes an approach which
detects moving and recently stopped vehicles using
the novel technique of background subtraction [1]
[2]. The algorithm is programmed, simulated and
tested in VHDL and then implemented on the FPGA
SPARTAN6 Board. The result of the algorithm is a
binary mask image of blobs representing the detected
objects. The background is updated slowly with
Selective and Non-selective algorithm. The use of
two updating blocks improves the sensitivity of the
algorithm. Also shadow detection block maintains
consistency of the algorithm by eliminating the error
introduced by shadow of an object.
1.1. About FPGA
Field Programmable Gate Arrays (FPGAs)
represent reconfigurable computing technology,
which is in some ways ideally suited for video
processing. Reconfigurable computers are processors
which can be programmed with a design, and then
reprogrammed (or reconfigured) with virtually
limitless designs as the designers need change. All of
the logic in an FPGA can be rewired, or reconfigured,
with a different design as often as the designer likes.
This type of architecture allows a large variety of
logic designs dependent on the processors resources),
which can be interchanged for a new design as soon
as the device can be reprogrammed. Engineers use a
hardware language such as VHDL, which allows for
a design methodology similar to software design.
www.ijera.com

This software view of hardware design allows for a
lower overall support cost and design abstraction.
1.2. About VHDL
VHDL is an acronym for Very High Speed
Integrated Circuit Hardware Description Language. It
is a hardware description language that can be used to
model a digital system at many levels of abstraction,
ranging from the algorithmic level to the gate level.
The VHDL language has constructs that enable you
to express the concurrent or sequential behaviour of a
digital system. It also allows you to model the system
as an interconnection of components. Test waveforms
can also be generated using the same constructs.
1.3. Algorithm
Each pixel is modified independently using
the statistical procedure of Gaussian distribution [3]
and the pixels of the moving object is detected using
the inequality mentioned below:
Where µt and σt are mean and standard deviation
matrices of Gaussian distribution for image pixel
intensities and constant k typically has a value
between 2 and 3.The updating background image is
calculated as shown by the following equations:
(2)

(3)
Where It-1 and µt-1 is the intensity of previous image
frame and previous frame and α is learning ratio.
1829 | P a g e

II.

www.ijera.com

NON-SELECTIVE BLOCK

The task of detecting the moving and
recently stopped vehicles is done by the non-selective
block. The equation for non-selective background
updating [2] [4] is given by:

Where It (x y) is the brightness of a pixel situated at
coordinates (x,y) of input monochrome image at the
time t; µN, t (x,y) the brightness of a pixel situated at
coordinates (x,y) of background image, updated nonselectively; δN1 = 2-5 = 0.03125 is a small constant
evaluated experimentally. It is assumed that the
brightness of the input image is in the range: It(x,y)
<0,255>.
The updating of
from (1) for non which is
selective model [2] is given by (5):

where
is experimentally evaluated to be
0.00390625 (i.e. 2-8). The values of µN,t and
are used to calculate the pixel values of mN
from equation (1).

(7)
where mVTS,t (x, y) = mV,t (x, y) ˅ mET,t (x, y)
mES,t
(x, y) ; S,t (x, y) – the brightness of a pixel at
coordinates (x, y) of background image updated
using selectivity; mV (x, y) – the element of the
detected vehicle mask image of value equal to 0 or 1,
where 1 denotes the detected moving objects. The
values of constants S1 and S2 were established
experimentally: S1 = 0.25, S2 = 0.03125 for It
x,
y)
. From (1), the image frame ms comes
out as shown below:

Figure 3: Showing ms(x,y)

IV.

Binary Mask Combination Block

The output from both selective and nonselective block is combined into a single binary
mask. A simple AND operation is not enough to
detect all the pixels. Hence a special combination of
AND and OR operations [5] are used to improve the
detection as shown (8):

Figure 2: Showing It (x,y) and mN (x,y)
respectively.

III.

Selective Block

This block is similar to the non-selective
block, but it depends on the final mask output [4] as
shown by (6):

Figure 4: Showing mB(x,y)

V.

(6)

www.ijera.com

Temporal and Spatial Edge Detection

During a dark scene, True Positive pixels [2]
may not be detected and a major portion or the car
may be neglected. So to improve the segmentation
quality, the TP pixels need to be increased. This is
done by using two type of edge detection: Temporal
and Spatial Edge Detection blocks.
1830 | P a g e

In temporal edge detection block, the edges are
detected by taking the difference of the current and
previous frame:
In spatial edge detection, the difference of current
image and background is taken as below:

VII.

www.ijera.com

Final Processing

The masks obtained by implementing
the previous algorithms are combined into a
single mask M_Temp1:

Temporal and spatial edge image mask mET and mES
respectively is given by the following equation:
where dil() and ero() denote 2x2 morphological
dilation and erosion operation [2] respectively.
The mask thus obtained usually contains False
negative and False positive pixels. Hence to improve
the shape of the blobs, the inbuilt opening block is
added to obtain the final frame of mask mV.
where θET and θES are constant threshold values.

Figure 7: Showing mV.

VIII.

Figure 5: Showing mET and mES respectively.

VI.

Shadow Detection

Shadows get detected and may be
misunderstood by the system to be an object itself.
Hence to remove any such discrepancies, the Shadow
detection technique is used. By comparing the
decrease in brightness [5] [6] [7], shadows are
detected by the following equation:
where α and β are constant coefficients: α=0.55,
β=0.95, both evaluated experimentally.

CONCLUSION

In this paper, the combined algorithm for
extracting moving objects from a stored video stream
is proposed. The processing steps were carefully
selected and adopted to provide simple and
straightforward realization in specialized hardware,
such as FPGA. A few novel ideas to enhance the
algorithm are also developed, increasing the
robustness and maintaining its simplicity for
hardware implementation. The proposed method of
background calculation, using running mode is very
fast and requires only basic operations. The novel
combination of masks from selective and nonselective background improves the detection quality.
Shadow Detection in various light conditions
improves the sensitivity. The application of Final
Processing significantly improves the final blob
mask. The test and simulation result proves the
usability of the presented idea for recognizing the
moving vehicles at low power consumption-the
system properly found almost all of the moving
vehicles during different climatic conditions.
IX.
ACKNOWLEDGEMENTS
We are thankful to Ms. Divya Shah (Asst.
Professor Electronics Dept.) for sharing their
profound knowledge in the field of VLSI. We also
want to thank college staff and family members for
their constant support.

REFERENCES
Figure 6: Showing mSH.

www.ijera.com

Journal Papers:
[1]. Sen-Ching S. Cheung and Chandrika
Kamath, “Robust Background Subtraction
With Foreground Validation For Urban
1831 | P a g e


[2].

[3].

[4].

[5].

[6].

[7].

www.ijera.com

Traffic Video” in Proc. SPIE Video
Communication and Image Processing,
2004,pp.881-892.
Marek Wójcikowski, Robert Zaglewski,
“FPGA-Based Real-Time Implementation of
Detection Algorithm for Automatic Traffic
Surveillance Sensor Network”.
A. Franois, G. Medioni, "Adaptive Color
Background Modeling for Real-Time
Segmentation of Video Streams", in Proc. of
Int. Conf. Imaging Science, Systems and
Technology, Las Vegas, NV, 1999, pp. 227232.
F. Porikli, Y. Ivanov, T. Haga, "Robust
abandoned object detection using dual
foregrounds", EURASIP J. Adv. Signal
Process,Jan.2008.
D. Duque, H. Santos, P. Cortez, "Moving
Object Detection Unaffected by Cast
Shadows,Highlights and Ghosts", in Proc.
IEEE Int. Conf. Image Processing, 2005, pp.
413-416.
R. Cucchiara, C. Granna. M. Piccardi, A.
Prati, S. Sirotti, "Improving Shadow
Suppression in Moving Object Detection
with HSV Color Information", in Proc. IEEE
Intell. Transp. Syst. Conf., Oakland, CA,
2001, pp. 334-339.
R. Cucchiara, C. Granna. M.Piccardi, A.
Prati, "Detecting Moving Objects, Ghosts
and Shadows in Video Streams", IEEE
Trans. Pattern Anal. Machine Intell., vol.
25, no. 10, pp. 1337-1342, Oct. 2003.

www.ijera.com

1832 | P a g e

Kq3518291832

More Related Content

What's hot (20)

Viewers also liked (18)

Similar to Kq3518291832 (20)

Recently uploaded (20)

Kq3518291832