4.

2 Edge Detection by Laplacian of Gaussian

a. LOG method at NF=0.05

Figure(4.11): shows the LOG method at 0.05 noise factor (salt & pepper).

b. LOG method at NF=0.1

Figure(4.12): shows the LOG method at 0.1 noise factor (salt & pepper).

1
c. LOG method at NF=0.3

Figure(4.13): shows the LOG method at 0.3 noise factor (salt & pepper).

d. LOG method at NF=0.5

Figure(4.14): shows the LOG method at 0.5 noise factor (salt & pepper).

2
e. LOG method at NF=0.7

Figure(4.15):shows the LOG method at 0.7 noise factor (salt & pepper).

f. LOG method at NF=0.9

Figure(4.16):shows the LOG method at 0.9 noise factor (salt & pepper).

3
3.5.2.1 Laplacian of Gaussian (LOG)
The Laplacian is a 2-D isotropic measure of the 2nd spatial derivative
of an image. The image highlights regions of rapid intensity change
and is therefore often used for edge detection. The Laplacian is often
applied to an image that has first been smoothed with something
approximating a Gaussian Smoothing filter in order to reduce its
sensitivity to noise. The operator normally takes a single gray level
image as input and produces another gray level image as output [15].

The general calling syntax for the LOG detector is

[g, t] = edge(f, 'log', T, sigma) (3.17)

The fundamental characteristics of the Laplacian of Gaussian edge

detector are:
1. The smoothing filter is a Gaussian.
2. The enhancement step is the second derivative (Laplacian in two
dimensions) .
3. The detection criterion is the presence of a zero crossing in the
second derivative with a corresponding large peak in the first
derivative.
4. The edge location can be estimated with sub pixel resolution using
linear interpolation.

In this approach,. an image should first be convolved with a Gaussian

filter. (We discuss Gaussian filtering in more detail in Section 3.5.)
This step smooth's an image and reduces noise. Isolated noise points
and small structures will be filtered out. Since the smoothing will
result in spreading of edges, the edge detector considers as edges
only those pixels that have locally maximum gradient. This is

4
achieved by using zero crossings of the second derivative. The
Laplacian is used as the approximation of the second derivative in
2-D because it is an isotropic operator. To avoid detection of
insignificant edges, only the zero crossings whose corresponding first
derivative is above some threshold are selected as edge points.

The output of the LOG operator, h(x, y), is obtained by the

convolution operation

h(x, y) = ∇2 [(g (x, y) * f (x, y)] (3.18)

Using the derivative rule for convolution,

h(x, y) = [∇2 g(x,y)] * f(x, y) (3.19)

where
𝑥2 +𝑦2
𝑥 2 +𝑦 2 −2𝜎 2 −( )
∇ g(x,y) = (
2
)𝑒 2𝜎2 (3.20)
𝜎4

Thus, the following two methods are mathematically equivalent:

1. Convolve the image with a Gaussian smoothing filter and compute
the Laplacian of the result.
2. Convolve the image with the linear filter that is the Laplacian of the
Gaussian filter.

Figure(3.5): The inverted Laplacian of Gaussian function, 𝜎 = 2, in one and two

dimensions.

5
 0 0 -1 0 0

0 -1 -2 -1 0

-1 -2 16 -2 -1

0 -1 -2 -2 0

0 0 -1 0 0
Figure(3.6): 5 x 5 Laplacian of Gaussian mask.