Read Me

You might also like

Download as txt, pdf, or txt
Download as txt, pdf, or txt
You are on page 1of 2

===================================================

Mean-Shift based Object Tracking #


Nguyen Duy Trung Duong #
Ho Hai Nam #
Le Hong Thang #
Class 09 ECE #
Center of Excellent #
Danang University of Technology #
March 2014 #
===================================================
We provide to use the tool for object tracking using mean-shift iteration
with the following m-function
1/ Density_estim: using kernel function to estimate the histogram of the input
patch

2/ Draw_target: in the frame I of the video, we use this function to draw the
tracking
rectangle from the necessary information in selecting tracking patch.
3/ GUI: the user interface equipped with the features of selecting the trackin
g video,
target selection, kernel radius selection, iteration number and similar function
threshold
4/ MeanShift_Tracking: implementing the Mean-Shift iteration algorithm into th
e code to
identify the next likelihood position for the target.
5/ MS_Tracking_EXAMPLE: An example using the video "ball.avi" using mean-shift
based
object tracking algorithm.
6/ Parzen_window: creating the corresponding kernel in a 2-D patch. The kernel
could be
Uniform, Triangular, Epanechnikov and Gaussian. Also, the gradient of the corres
ponding
kernel.
7/ Select_patch: providing the first frame from the video so that we can selec
t the
tracking patch for the target. As a result, we receive the height, width of the
tracking
rectangle.
8/ Simil_func: determining the similarity of the two density distributions and
deriving
the weight in the similarity function resulted from Taylor expansion.
9/ Target_Loss_Dialog_Box: the dialog box that inform the tracking object has
disappeared
from the current frame.
################################################################################
###########
DEMO
____
Run the "MS_Tracking_EXAMPLE.m" and then draw the tracking rectangle to track th
e target
OR
Run the "GUI.m" demo and chose the desire video for tracking. Similarly, draw th
e tracking
rectangle and chose the type of kerne and its radius. Furthermore, you need to a
djust
the parameter of the algorithm ( 0< threshold < 1, and the number of iteration,
0< iteration <1)
==> Note that in the "MS_Tracking_EXAMPLE.m", the kernel type, radius and algori
thm's
parameters are set in the code:
line 45: kernel type =... ('Uniform','Triangular','Epanechnikov' and 'Gaussian')
line 46: radius = ... (0.00~1.00)
line 41: f_thresh = ...(0.00~1.00)
line 43: max_it = ...(max is 15)

You might also like