Welcome to Scribd, the world's digital library. Read, publish, and share books and documents. See more
Download
Standard view
Full view
of .
Save to My Library
Look up keyword
Like this
1Activity
0 of .
Results for:
No results containing your search query
P. 1
A Novel Approach for Object Detection and Tracking using IFL Algorithm

A Novel Approach for Object Detection and Tracking using IFL Algorithm

Ratings: (0)|Views: 446|Likes:
Published by ijcsis
This paper is an innovative attempt has been made using Attanassov’s Intuitionistic fuzzy set theory for tracking moving objects in video. The main focus of this proposed work is taking an account for handling uncertainty in assignment of membership degree known as hesitation degree using Intuitionistic fuzzy. Many algorithms have been developed to reduce the computational complexity of movement vector evaluation. In this paper we propose to implement Intuitionistic logic based block Matching Algorithm termed as BMIFL to overcome the computational complexity. In this proposed methodology feature extraction is performed using 2Dfilter, segmentation using approximate median and object detection is done using our proposed algorithm Intuitionistic fuzzy. The results obtained clearly shows that our algorithm performs better than fuzzy logic based three Step Search algorithm.
This paper is an innovative attempt has been made using Attanassov’s Intuitionistic fuzzy set theory for tracking moving objects in video. The main focus of this proposed work is taking an account for handling uncertainty in assignment of membership degree known as hesitation degree using Intuitionistic fuzzy. Many algorithms have been developed to reduce the computational complexity of movement vector evaluation. In this paper we propose to implement Intuitionistic logic based block Matching Algorithm termed as BMIFL to overcome the computational complexity. In this proposed methodology feature extraction is performed using 2Dfilter, segmentation using approximate median and object detection is done using our proposed algorithm Intuitionistic fuzzy. The results obtained clearly shows that our algorithm performs better than fuzzy logic based three Step Search algorithm.

More info:

Categories:Types, Research
Published by: ijcsis on May 08, 2013
Copyright:Attribution Non-commercial

Availability:

Read on Scribd mobile: iPhone, iPad and Android.
download as PDF, TXT or read online from Scribd
See more
See less

05/08/2013

pdf

text

original

 
 
A Novel Approach For Object Detection andTrackingusing IFL Algorithm
R.Revathi
Research Scholar,Dept. of Computer ScienceKarpagam UniversityCoimbatore,India
M.Hemalatha
Dept. of Computer ScienceKarpagam UniversityCoimbatore,India
 Abstract
—This paper is an innovative attempt has been madeusing Attanassov’s Intuitionistic fuzzy set theory for trackingmoving objects in video. The main focus of this proposed work istaking an account for handling uncertainty in assignment of membership degree known as hesitation degree usingIntuitionistic fuzzy. Many algorithms have been developed toreduce the computational complexity of movement vectorevaluation. In this paper we propose to implement Intuitionisticlogic based block Matching Algorithm termed as BMIFL toovercome the computational complexity. In this proposedmethodology feature extraction is performed using 2Dfilter,segmentation using approximate median and object detection isdone using our proposed algorithm Intuitionistic fuzzy. Theresults obtained clearly shows that our algorithm performs betterthan fuzzy logic based three Step Search algorithm
 Keywords-component; Noise filtering,Segmentation,ObjectTracking and detection,Fuzzy Logic.
I.
 
I
 NTRODUCTION
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, securityand surveillance, video communication and compression,augmented reality, traffic control, medical imaging [1] andvideo editing.
[2][3]
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 touse object recognition techniques for tracking [4]. Theassociation can be especially difficult when the objects aremoving fast relative to the frame rate. Another situation thatincreases the complexity of the problem is when the trackedobject changes orientation over time. [3].II.
 
ELATE
D
WORKS
 Fuzzy controller system has been suggested which created atime, according to the 2 or 3 arrival
 
 parameters and their evaluation. This created time is related to the increasing of time needed when vehicles cross the junction. Shilpa et al.(2008) divided a street into 3 longitudinal traffic lanes throughcamera sensor and image processing. A crossing chance is provided in each lane. An operation is a function performedaccording to phases. Khiang and Khalid et al. (1996)simulated traffic junction on two kinds of controller system(ordinary and fuzzy), according to cases such as waiting time,traffic density, cost etc. Barzegar et al. (2011) introduced thesimulation of traffic light controller by Fuzzy Petri net throughimplemented operations.An intelligent traffic light monitoring system using anadaptive associative memory was designed by Abdul Kareemand Jantan (2011). The research was motivated by the need toreduce the unnecessary long waiting times for vehicles atregular traffic lights in urban area with 'fixed cycle' protocol.To improve the traffic light configuration, the paper proposedmonitoring system, which will be able to determine threestreet cases (empty street case, normal street case and crowdedstreet case) by using small associative memory. Theexperiments presented promising results when the proposedapproach was applied by using a program to monitor oneintersection in Penang Island in Malaysia. The program coulddetermine all street cases with different weather conditionsdepending on the stream of images, which are extracted fromthe streets video cameras [8]A distributed, knowledge-based system for real-time andtraffic-adaptive control of traffic signals was described byFindler and et al (1997). The system was a learning system intwo processes: the first process optimized the control of steady-state traffic at a single intersection and over a network of streets while the second stage of learning dealt with predictive/reactive control in responding to sudden changes intraffic patterns [9]. GiYoung et al., (2001) believed that electrosensitive traffic lights had better efficiency than fixed presettraffic signal cycles because they were able to extend or shorten the signal cycle when the number of vehicles increasesor decreases suddenly. Their work was centred on creating anoptimal traffic signal using fuzzy control. Fuzzy membershipfunction values between 0 and 1 were used to estimate theuncertain length of a vehicle, vehicle speed and width of aroad and different kinds of conditions such as car type, speed,delay in starting time and the volume of cars in traffic werestored [10]. A framework for a dynamic and automatic trafficlight control expert system was proposed by [11]. The modeladopted inter-arrival time and inter-departure time to simulatethe arrival and leaving number of cars on roads. Knowledge base system and rules were used by the model and RFID were
(IJCSIS) International Journal of Computer Science and Information Security,Vol. 11, No. 4, April 2013102http://sites.google.com/site/ijcsis/ISSN 1947-5500
 
 
deployed to collect road traffic data. This model was able tomake decisions that were required to control traffic atintersections depending on the traffic light data collected bythe RFID reader. A paper by Tan et al., (1996) described thedesign and implementation of an intelligent traffic lightscontroller based on fuzzy logic technology. The researchersdeveloped a software to simulate the situation of an isolatedtraffic junction based on this technology. Their system washighly graphical in nature, used the Windows system andallowed simulation of different traffic conditions at the junction. The system made comparisons the fuzzy logiccontroller and a conventional fixed-time controller; and thesimulation results showed that the fuzzy logic controller had better performance and was more cost effective [12].Research efforts in traffic engineering studies yielded thequeue traffic light model in which vehicles arrive at anintersection controlled by a traffic light and form a queue.Several research efforts developed different techniquestailored towards the evaluation of the lengths of the queue ineach lane on street width and the number of vehicles that areexpected at a given time of day. The efficiency of the trafficlight in the queue model however, was affected by theoccurrence of unexpected events such as the break-down of avehicle or road traffic accidents thereby causing disruption tothe flow of vehicles. Among those techniques based on thequeue model was a queue detection algorithm proposed by[13]. The algorithm consisted of motion detection and vehicledetection operations, both of which were based on extractingthe edges of the scene to reduce the effects of variations inlighting conditions. A decentralized control model wasdescribed Jin & Ozguner (1999). This model was acombination of multi-destination routing and real time trafficlight control based on a concept of costto- go to differentdestinations [14]. A believe that electronic traffic signal isexpected to augment the traditional traffic light system infuture intelligent transportation environments because it hasthe advantage of being easily visible to machines was propagated by Huang and Miller (2004).Their work presented a basic electronic traffic signaling protocol framework and two of its derivatives, a reliable protocol for intersection traffic signals and one for stop signsignals. These protocols enabled recipient vehicles to robustlydifferentiate the signal’s designated directions despite of  potential threats (confusions) caused by reflections. Theauthors also demonstrated how to use one of the protocols toconstruct a sample application: a red- light alert system andalso raised the issue of potential inconsistency threats caused by the uncertainty of location system being used and discussmeans to handle them [15]. Di Febbraro el al (2004) showedthat Petri net (PN) models can be applied to traffic control.The researchers provided a modular representation of urbantraffic systems regulated by signalized intersections andconsidered such systems to be composed of elementarystructural components; namely, intersections and roadstretches, the movement of vehicles in the traffic network wasdescribed with a microscopic representation and was realizedvia timed PNs. An interesting feature of the model was the possibility of representing the offsets among different trafficlight cycles as embedded in the structure of the model itself [16]. Nagel and Schreckenberg (1992) described a Cellular Automata model for traffic simulation. At each discrete time-step, vehicles increase their speed by a certain amount untilthey reach their maximum velocity. In case of a slower moving vehicle ahead, the speed will be decreased to avoidcollision. Some randomness is introduced by adding for eachvehicle a small chance of slowing down [17].The experiences of building a traffic light controller using asimple predictor was described by Tavladakis (1999).Measurements taken during the current cycle were used to testseveral possible settings for the next cycle, and the settingresulting in the least amount of queued vehicles was executed.The system was highly adaptive, however as it only uses dataof one cycle and could not handle strong fluctuations in trafficflow well [18]. Chattarajet al., (2008) proposed a novelarchitecture for creating Intelligent Systems for controllingroad traffic. Their system was based on the principle of the useof Radio Frequency Identification (RFID) tracking of vehicles.This architecture can be used in places where RFID tagging of vehicles is compulsory and the efficiency of the system lied inthe fact that it operated traffic signals based on the currentsituation of vehicular volume in different directions of a roadcrossing and not on pre-assigned times [19].III.
 
O
BJECT
T
RACKING PROPOSED
M
ETHODOLOGY
 IV.
 
PHASES
 
USED
 
IN
 
OBJECT
 
TRACKING
 A.
 
 Noise:
The most significant stages in image processingapplications are the noise filtering. The importance of image sequence processing is regularly increasing with theever use of digital television and video systems inconsumer, commercial, medical, and communicational
(IJCSIS) International Journal of Computer Science and Information Security,Vol. 11, No. 4, April 2013103http://sites.google.com/site/ijcsis/ISSN 1947-5500
 
 
applications. Image filtering is not only used to improvethe image quality but also is used as a preprocessing stagein many applications including image encoding, patternrecognition, image compression and target tracking, toname a few. This preprocessing stage is essential in mostof the image-processing algorithm and improper noisefiltering may result in inappropriate or even false outcome.Different methods have been proposed for the purpose of noise filtering. [20].1.
 
Select three videos which contain three differentnoises like -Salt and pepper noise/ Gaussian noise/periodic noise.2.
 
Convert videos to Frames.3.
 
Apply various filters in the noise generated frames4.
 
Identify the best suited filter using the PSNR andMSE
5.
 
Use the resultant frames for further processing
.
From the results obtained we conclude that with threedifferent noises salt and pepper noise, Gaussian noise and periodic noise applied for denoising of the spatial video produces variant results over different filtered techniques.From the results obtained using various filtering techniques itis observed that for salt and pepper noise median and rank order filter works better than other techniques. In case of Gaussian noise Weiner and rank order filter works fine. For Periodic noise 2D filter works better than other filters.
 B.
 
Segmentation:
Segmentation is the method of partitioning a digital imageinto multiple segments (sets of pixels, also known as super  pixels). The goal of segmentation is to make simpler and/or change the representation of an image into something that ismore meaningful and easier to analyze.[21] Imagesegmentation is characteristically used to trace objects and boundaries (lines, curves, etc.) in images.
1)
 
 Approximate median segmentation
Approximate median method uses a recursive method for estimating a background model. Each pixel in the backgroundmodel is compared to the corresponding pixel in the currentframe, to be incremented by one if the new pixel is larger thanthe background pixel or decremented by one if smaller. A pixel in the background model effectively converges to a valuewhere half of the incoming pixels are larger than and half aresmaller than its value. This value is known as the median.
a)
 
Process:
1. Assign the variables move to the input video, n frames tothe number of frames, set the threshold value to 25 and movethe frames one by one to the n (i).cdata.2. Read the 1st background frame as bg=n(1).cdata andconvert it into gray scale3. set the frame size variables fr-size to the size of the background frame and width and height corresponding to thefr_size.4. convert all the frames into grayscale and type cast theoperands as double to avoid negative overflowUsing fr_diff = abs (double (fr_bw) - double(bg_bw));5. If fr_diff (frame difference) of the considered frame isgreater than the threshold pixel in the foreground thenincrement background value else decrement the background pixel value.6. Continue step 5 for all width varying from 1 and heightvarying from 1.7. Display the result using plot and imshow frame.8. If needed save the output as movie.
C.
 
Feature extraction
The feature is defined as a function of one or moremeasurements, each of which specifies some quantifiable property of an object, and is computed such that it quantifiessome significant characteristics of the object. [22].Feature Extraction plays a major role to detect the movingobjects in sequence of frames. Every object has a specificfeature like color or shape. In a sequence of frames, any one of the feature is used to detect the objects in the frame. [23]
1)
 
 Bounding Box with Color Feature
 
If the segmentation is performed using frame difference, theresidual image is visualized with rectangular bounding boxwith the dimensions of the object produced from residualimage. For a given image, a scan is performed where theintensity values of the image are more than limit (depends onthe assigned value, for accurate assign maximum). In thisFeatures is extracted by colour and here the intensity valuedescribes the color. The pixel values from the first hit of theintensity values from top, bottom, left and right are stored. Byusing this dimension values a rectangular bounding box is plotted within the limits of the values produced.[23]
a)
 
 Algorithm for Bounding Box:
1.
 
Read the Image difference2.
 
For (pres pos=int value: final Value)of y resolution3.
 
For (pres pos=int value: final Value)of x resolutiona.
 
Calc the sharp change in intensity of image fromtop and bottom b.
 
Store the values in an array4.
 
Height of the bounding box is = bottom value – topvalue5.
 
For (pres pos=int value: final Value)of x resolution
(IJCSIS) International Journal of Computer Science and Information Security,Vol. 11, No. 4, April 2013104http://sites.google.com/site/ijcsis/ISSN 1947-5500

You're Reading a Free Preview

Download
/*********** DO NOT ALTER ANYTHING BELOW THIS LINE ! ************/ var s_code=s.t();if(s_code)document.write(s_code)//-->