Saturday, October 25, 2008

Video Object Tracking

by Your Name 0 comments

Tag


Share this post:
Design Float
StumbleUpon
Reddit

Abstract  –  The  advance  of  technology  makes  video  acquisition  devices  better  and  less  costly,  thereby  increasing  the number of applications  that can  effectively  utilize  digital  video.  Compared  to  still  images,  video sequences  provide  more  information  about  how  objects change over  time. Different applications can be developed through  detection  of  moving  objects  in  video.  We  have identified  two  major  applications  
of  the  Video  Object tracking algorithm. The algorithm can be used to estimate the  speed  of  the  moving  object  and  can  also  be  used  to optimize  the  storage  space  of  the  recorded  video  in  the surveillance system.  
 
Index Terms – Segmentation, optimization,  speed  estimation, sobel  algorithm,  rectangularization,  static  camera,  object tracking. 
INTRODUCTION 
Video tracking is the process of locating a moving object (or  several  ones)  in  time  using  a  camera.  An  algorithm analyses  the video frames and outputs  the location of moving 
targets within the video frame. The main difficulty in video tracking is to associate target 
locations  in  consecutive  video frames,  especially  when  the objects are moving fast relative to the frame rate. Here, video tracking  systems  usually  employ  a  motion  model  which 
describes  how  the  image  of  the  target  might  change  for different possible motions of the object to track. 
The Video Object Tracking (VOT) system operates in the controlled  campus  environment  of  Infosys  Technologies Limited. The system  takes  the video as  input  from  the single static  camera  and  does  the  processing.  It  starts  recording  the video after the motion is detected and also computes the speed of moving object. 
 PROBLEM DEFINITION 
Moving  objects  need  to  be  tracked  and  analyzed  in videos.  We  are  working  in  a  controlled  campus  like environment where a single static camera is used to record the 
objects motion.  This  reduces  the  complexities  involved with moving  or  multiple  cameras.  After  the  object’s  motion  is detected  it will be used  to develop different applications  like 
peed estimation and storage space optimization. 
RELATED WORK 
 There have been  few  algorithms and  techniques  used  to track  objects  through  video  devices.  Like  tracking  moving objects  through video camera, semi-automatic object  tracking 
in video etc. Each of the methods work on different algorithms to  track  the  objects.  Every  method  has  its  advantages  and disadvantages. Listing the techniques briefly – 
I. Moving Object Tracking in Video [1] 
The algorithm first separates the moving objects from the background  in  each  frame.  Then,  four  sets  of  variables  are computed  based  on  the  positions,  the  sizes,  the  grayscale 
distributions  and  the  presence  of  textures  of  the  objects.  A rule-based method  is  developed  to  track  the  objects  between frames, based on the values of the variables. 
II. Real-Time Object tracking from a moving camera [2] 
The algorithm is based on describing  the displacement of a point as a probability distribution over a matrix of possible displacements. A small set of randomly selected points is used 
to  compute  the  registration  parameters.  Moving  object detection  is  based  on  the  consistency  of  the  probabilistic displacement of image points with the global image motion. 
III. Semi-auto object tracking in video sequences [3]  
A method is presented for semi-automatic object tracking in  video  sequences  using multiple  features  and  a method  for probabilistic  relaxation  to  improve  the  tracking  results 
producing smooth and accurate tracked borders. Starting from a  given  initial  position  of  the  object  in  the  first  frame  the proposed  method  automatically  tracks  the  object  in  the 
sequence  modeling  the  a  posteriori  probabilities  of  a  set  of features such as color, position and motion, depth, etc. 
 

Comments 0 comments

Subscribe feeds via e-mail
Subscribe in your preferred RSS reader

Subscribe feeds rss Recent Entries

Advertise on this site Sponsored links

Categories

Subscribe feeds rss Recent Comments

Technorati

Technorati
My authority on technorati
Add this blog to your faves