You are on page 1of 26

Submitted by,

y07it051
y07it062
y07it060
1
Contour detection using Hough Transform

Contour detection using Active Snakes Contours

Contour detection using Geometric Active


Contours (GAC)

2
Hough Transform
Definition:

Extracts geometric primitives from raster


images.

Identifies lines in an image, positions of arbitrary


shapes like circle, ellipse

Importance of Hough Transform


3
Basic idea
Each white point if considered in
isolation could lie on an infinite
number of straight lines

In the Hough transform each point


votes for every line it could be on

The lines with the most votes win

4
Representation of lines
Any line can be represented by two numbers

Here we will represent the yellow line by (w 

In other words we define it using


- a line from an agreed origin
- of length w
- at angle  to the horizontal

w  x cos(  y sin( 


5
Algorithm:
Create and w for all possible lines
Create an array A indexed by and w
for each point (x,y)
for each angle 
w = x*cos()+ y*sin()
A[,w] = A[,w]+1
end
end

6
Representation of multiple lines

Single point representation

7
Two points representation
Active Snake Contours
Definition: an energy-minimizing spline that
represents an object boundary or some other salient
image feature as a parametric curve.

It was first introduced in 1987 by Kass et al

The contour is defined in the (x, y) plane of an image as


a parametric curve

8
Modeling
Contour is said to possess an energy (Esnake) which is defined as
the sum of the three energy terms.

Elastic energy Bending energy

dv ( s )
1 vs 
Eint  Eelastic  Ebending    | vs |2  | vss |2 )ds ds
s
2
1 1 1
   2
Eelastic   ( s ) | vs |2 ds Ebending
2
( s ) | v ss | ds
2s s

9
External energy
It is derived from the image.

Define a function Eimage(x , y) so that it takes on its


smaller values at the features of interest, such as
boundaries.
Eext   Eimage (v( s ))ds
s

10
Image forces
• The image forces E image are derived from the image data over
which the snake lies
• Three important features the snake can be attracted to are line,
edge and termination functions. The total image energy can be
expressed as a weighted combination of the three.

Eimage   line E line   edge E edge   term E term


The simplest useful image functional is the image intensity
Eline =I (x,y)
E edge = square of intensity variation
E term =curvature of lines
11
Example

12
MOTION TRACKING

13
Geometric Active
Contours
Level set method is used instead of parameters to trace
multiple objects more specifically .

The level set method is a numerical technique for tracking


interfaces and shapes.

14
Level set method
Advantage:
 perform numerical computations
 Make it very easy to follow shapes that change topology.

the level set method represents a  closed curve, Γ using an auxiliary


function ф, called the level set function.
  Γ is represented as the zero level set of ф by
Γ={(x , y)|ф(x , y)=0}

 If the curve Γ moves in the normal direction with a speed v, then the level
set function  satisfies the 
level set equation:
∂ф /∂t=v| ф|

15
Approach:
 Curve deformation velocity, v regularity of the curve

shrinks or expands the curve


towards the boundary.

 Given a 2-D scalar function, ф(x , t) geometric active contour formula is:

Level set operator


constant
∂ф /∂t=c(к+Ѵ0) ф

the use of curvature к has the effect of smoothing the contour, while
 Ѵ0 has the effect of shrinking or expanding contour at a constant speed.

 c(к+Ѵ0) determines the overall evolution speed of level sets of ф(x , t)

16
where, c=c(x)= 1 / [1+| |G σ (x) * I(x)]
here,
 x is any location in the image, I(x) and

 G σ (x) is a 2-D Gaussian function with standard deviation σ


- filtering noise
 is the gradient operator- measure of intensity variation

 * is the 2-D image convolution operator

Advantage:
Several objects can be detected simultaneously without previous knowledge
of their exact number in the scene

17
x − y plane

level set function determining


the shape of the object

18
19
Initial contour

20
21
22
23
Detecting multiple objects
using geometric contours

amoeba

24
Summary
Hough transform – cannot detect shapes other than
lines and circles

Active Snakes Contours – selection of parameters


- multiple objects

Geometric Active Contours


– overcomes above disadvantages using level sets
- detecting multiple objects

25
Thank You…

1 26

You might also like