September 2016, Volume 3, Issue 9 JETIR (ISSN-2349-5162)

JETIR1609005 Journal of Emerging Technologies and Innovative Research (JETIR) www.jetir.org 26

IMPLEMENTATION OF THE GAUSSIAN MIXTURE MODEL ALGORITHM FOR FOREGROUND DETECTION USING SEGMENTATION AND

THRESHOLDIND FOR HIGH DEFINITION VIDEO SIGNAL

1Mr. Sateesh Kumar, 2Mr. Rupesh Mahamune 1M. Tech. Scholar (Digital Electronics), 2Asst. Professor ET&T Department

1,2Rungta College of Engineering & Technology, Bhilai, CSVTU,

Abstract- Foreground detection is very critical in video signal. For foreground detection we take video signal from the static camera. In

foreground detection we follow these steps pre-processing, background modelling and foreground detection. In preprocessing we removes the

noise in background modelling we model the background model, in foreground detection in this paper we detect the foreground by the

comparative analysis of number of Gaussian frame. Result we compare by the help of comparative analysis of number of Gaussian frame

and percentage foreground detection.

Key words: - foreground, background, clustering, GMM, threshold, MSE, PSNR

I. INTRODUCTION

According to relativity theory, we assume that time axis is constant and when position of any object is changed with respect to time then it is

called moving object. In video signal position of moving object is changed with respect to time. By the help of background subtraction algorithm

technique, we detect the moving object in video signals. Detection of moving object is very critical and challenging task in real-time video

signal. This striving becomes more critical in indoor and outdoor video signal under these circumstances such as illumination changed or steady

and dynamic weather condition.

II. METHODOLOGY FOR MOVING OBJECT DETECTION AND RESULT

For moving object this steps are follows. Block diagram for foreground is the given below fig 1. We first take the input in the form of video

signal then after we break the video signal into image frame. Then after we take the frame from the broken frame of video signal.

Processing

Foreground

Background Modelling

Foreground

Detection

Fig 1: Block Diagram of Methodology

Input Pre-

September 2016, Volume 3, Issue 9 JETIR (ISSN-2349-5162)

JETIR1609005 Journal of Emerging Technologies and Innovative Research (JETIR) www.jetir.org 27

Pre-processing-

In pre-processing we remove the noise from the image. In this project we take the two type of noise which is present in MATLAB library this

noise Gaussian noise and Salt and Pepper noise. Fig 2 shows that the original image and sharpened image. For filtering we take the mean filter.

Fig 2: original image and sharpened image

Background Modelling: Background modelling plays very vital role in background subtraction algorithm. Background estimation: The absolute background subtracting estimation by the help of this equation

IF D=| C − B | >T

O =1 (OBJECT) (1)

ELSE

O = 0 (BACKGROUND)

Where C = pixel value of current images.

B = pixel value of background images.

D =absolute difference of current and background image

O = binary difference images.

Threshold error is produced between current pixel and background pixel.

Foreground Detection: In foreground detection in this project we segmented image by the help of segmented image. Group of pixel in the image is defined as region.

Region of pixel may be any shape such as circle; ellipse, polygon etc. when area of interest in the images does not extracted exactly then we use

segmentation technique. Segmentation has two objectives: The first objective is to decompose the image into parts for further analysis. Figure 3

show that the original image and detected object is given below.

Fig 3 Original Image and Detected Object

September 2016, Volume 3, Issue 9 JETIR (ISSN-2349-5162)

JETIR1609005 Journal of Emerging Technologies and Innovative Research (JETIR) www.jetir.org 28

In background subtraction algorithm we compare two consecutive frames. Advantage of Background subtraction algorithm is does not

previous knowledge of the frame it just take two consecutive frames so background subtraction algorithm is best method other than. After

performing background subtraction algorithm we find the foreground which is given below figure 4. By the help of background subtraction we

find the object with the help of MATLAB fig 6 shows foreground

Fig 4 Foreground after Background Subtraction Algorithm

Filtering – Noise is unwanted signal which creates disturbance in input signal. In this paper we take the mean filter for removing noise in this paper we

take the two types of noise Gaussian noise and salt and pepper noise. Fig 5 shows that the clean foreground.

Fig 5 Clean Foreground by Background Subtraction

We take the image at different Gaussian frame figure 6 shows that the image at different Gaussian frame as given below figure 6.

Fig 6 foreground by comparative analysis Num training frame

September 2016, Volume 3, Issue 9 JETIR (ISSN-2349-5162)

JETIR1609005 Journal of Emerging Technologies and Innovative Research (JETIR) www.jetir.org 29

III. PERFORMANCE EVOLUTION

We compare the result by the help of number of Gaussian frame after experiment we find the when number of Gaussian frame is increased

then percentage of foreground detection is increased comparative analysis is given below table 1.

Table 1. Comparative Analysis at Different Gaussian Value

Number of Gaussian frame Percentage Foreground Detection

10 77.11

20 86.74

30 87.89

40 89.07

50 89.43

60 90.02

When we compare our result of proposed algorithm for foreground detection with respect to [17] then we find that the percentage foreground

detection is increased. In given below fig 7 shows that image in which we compare result and table 2 shows that the comparative analysis of

percentage foreground detection.

Table 2 Comparative Analysis of Foreground Detection

Y Cb Cr feature HSV feature RGB feature multi features Proposed

algorithm

Similarity value 70 % 25 % 60 % 90 % 91.76%

ACKNOWLEDGEMENT We are thankful to Rungta College of Engineering and Technology, Bhilai, C.G., India, for providing the laboratory facilities.

REFERENCES

[1] Chuanxu Wang Weijuan Zhang “A Robust Algorithm for Shadow Removal of Foreground Detection in Video Surveillance” 2009 Asia-

Pacific Conference on Information Processing.

[2] T. Horprasert, David Harwood Larry S. Davis A Statistical Approach for Real-time Robust Background Subtraction and Shadow

Detection. Computer Vision Laboratory University of Maryland

[3] Chris Stauffer W.E.L Grimson “Adaptive background mixture models for real-time tracking ” The Artificial Intelligence Laboratory

Massachusetts Institute of Technology.

[4] P KaewTraKulPong1,2 and R Bowden1 “An Adaptive Visual System for Tracking Low Resolution Color Targets” pp. 243-252

[5] Minghua Shi and Amine Bermak, “An Efficient Real Time Moving Object Detection Method for Video Surveillance System” IEEE

TRANSACTIONS ON VERY LARGE SCALE INTEGRATION (VLSI) SYSTEMS, VOL. 14, NO. 9, pp. 962-974 SEPTEMBER

2006

[6] S.C. S. Cheung and C. Kamath, “Robust Techniques For Background Subtraction In Urban Traffic Video”, IS&T/SPIE's Symposium on

Electronic Imaging San Jose, CA, United States pp. 183-192 January 18-22, January 2004 San Jose, CA, United States .

[7] Nan Lu, Jihong Wang, Li Yang, Henry Wu “Motion Detection Based on Accumulative Optical Flow and Double Background Filtering”

Proceedings of the World Congress on Engineering 2007 Vol I WCE 2007, July 2 - 4, 2007, London, U.K.

[8] Shao-Yi Chien, Shyh-Yih Ma, and Liang-Gee Chen, “Efficient Moving Object Segmentation Algorithm using Background registration

Technique” IEEE Transactions on Circuits and Systems for Video Technology, VOL. 12, NO. 7, pp- 577-586, JULY 2002.

[9] Nawaz, M. Cosmas, J. ; Lazaridis, P.I. ; Zaharis, Z.D. ; Yue Zhang ; Mohib, H. “Precise Foreground Detection Algorithm using Motion

Estimation Minima and Maxima inside the Foreground Object” Broadcasting, IEEE Transactions on (Volume:59 , Issue: 4 ) , pp-725-

731 ,21 October 2013.

[10] Shui-gen Wei a, Lei Yang a, Zhen Chen a ,Zhen-feng Liu “Motion Detection Based on Optical Flow and Self Adaptive Threshold

Segmentation” pp-3471 – 3476 Advanced in Control Engineering and Information Science ELSEVIER, Procedia Engineering 15 (2011)

[11] Katharina Quast, Matthias Obermann and Andr´e Kaup “Real Time Moving Object Detection in Videos Sequence Using Saptio

Temporal Adaptive Gaussian Mixture Model” pp- 413-418VISAPP 2010 - International Conference on Computer Vision Theory and

Applications.

[12] Vesna ZELJKOVIĆ, Željen TRPOVSKI, Vojin ŠENK “ New Algorithm for Moving Object Detection” pp 117-132 Yugoslav Journal

of Operations Research 14 (2004), Number 1,

September 2016, Volume 3, Issue 9 JETIR (ISSN-2349-5162)

JETIR1609005 Journal of Emerging Technologies and Innovative Research (JETIR) www.jetir.org 30

[13] Darren E. Butler V.Michael Bove Jr. “ Real –Time Adaptive Foreground / Background Segmentation”EURASIP Journal on Applied

Signal Processing 2005:14, 2 pp. 292–23042005 Hindawi Publishing Corporation.

[14] Abhishek Kumar Chauhan, Prashant KrishanMoving “Object Tracking using Gaussian Mixture Model and Optical Flow” Volume 3,

Issue 4, April 2013 pp. 243-246International Journal of Advanced Research in Computer Science and Software Engineering

[15] Mary Rose Simon, Jincy Fernandez “Survey on Various Technique in Region Based Foreground Detection” Volume-2, Issue-6, June-

2014 pp. 83-86 International Journal of Advanced Computational Engineering and Networking, ISSN: 2320-2106,

[16] Chunsheng FANG (Victor) “Gaussian Mixture Model background modeling for video Manual version 1.0” Chunsheng Fang, Univ. of

Cincinnati, 2008.

[17] Dr. Emad Kadum Jabbar, Dr. Ahmed Tariq Sadiq, Nuha Jameel Ibrahim “Foreground Detection By Using Multi Features” IOSR

Journal of Computer Engineering (IOSR-JCE) E-ISSN: 2278-0661, P- ISSN: 2278-8727Volume 16, Issue 2, Ver. X (Mar-Apr. 2014),

PP 08-12 www.iosrjournals.Org.