Scribd
Upload
36 views5 pages

Computer Vision & Digital Image Processing

The document discusses image segmentation using thresholding. It explains that thresholding involves selecting a threshold value T that separates the bimodal histogram of an image into light objects and a dark background. Thresholding can be global, local, or dynamic depending on whether T depends on pixel grayscale values alone or also on neighboring pixel properties or spatial location. Boundary-based thresholding aims to improve the histogram by considering only pixels near object boundaries, making the peaks more symmetric and valleys deeper. Gradients and Laplacians can identify boundary pixels and determine what side they fall on to create a three-level boundary characteristic image for thresholding.

Uploaded by

Siva Dharshan Valuvan
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
36 views5 pages

Computer Vision & Digital Image Processing

The document discusses image segmentation using thresholding. It explains that thresholding involves selecting a threshold value T that separates the bimodal histogram of an image into light objects and a dark background. Thresholding can be global, local, or dynamic depending on whether T depends on pixel grayscale values alone or also on neighboring pixel properties or spatial location. Boundary-based thresholding aims to improve the histogram by considering only pixels near object boundaries, making the peaks more symmetric and valleys deeper. Gradients and Laplacians can identify boundary pixels and determine what side they fall on to create a three-level boundary characteristic image for thresholding.

Uploaded by

Siva Dharshan Valuvan
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 5

Computer Vision &

Digital Image Processing

Image Segmentation: Thresholding

Electrical & Computer Engineering Dr. D. J. Jackson Lecture 18-1

Image segmentation: thresholding

• Suppose an image f(x,y) is composed of several light


objects on a dark background.
• The histogram for such an image may look like the
following: showing two dominate modes
• An obvious way to extract object information is to select a
threshold T that separates the two modes

T
Electrical & Computer Engineering Dr. D. J. Jackson Lecture 18-2
Thresholding (continued)

• Suppose several objects with differing gray levels (with a


dark background) comprise the image
• An object may be classified as belonging to one object
class if T1<f(x,y) ≤T2, to a second class if f(x,y)>T2 or to the
background if f(x,y) ≤T1
• This, however, is generally less reliable than single level
thresholding

T1 T2
Electrical & Computer Engineering Dr. D. J. Jackson Lecture 18-3

Thresholding (continued)

• Thresholding may be viewed as an operation that tests


against a given function of the form
T = T [ x, y, p ( x, y ), f ( x, y )]
• where f(x,y) is the gray level of point (x,y) and p(x,y) is some
local property of the point -- the average gray level of a
neighborhood around (x,y)
• The thresholded image is given by
⎧1 if f(x, y) > T
g ( x, y ) = ⎨
⎩0 if f(x, y) ≤ T
• Pixels labeled 1 (or any other convenient gray level value)
correspond to objects

Electrical & Computer Engineering Dr. D. J. Jackson Lecture 18-4


Thresholding (continued)

• When T depends only on f(x,y) the threshold is called global


• If T depends on f(x,y) and p(x,y) the threshold is local
• If, in addition, T depends on the spatial coordinates (x,y), the
threshold is called dynamic
• For example, a local threshold may be used if certain
information about the nature of the objects in the image is
known a priori
• A dynamic threshold may be used in the case where object
illumination is non-uniform

Electrical & Computer Engineering Dr. D. J. Jackson Lecture 18-5

Thresholding based on boundaries

• Important aspect of threshold selection: the ability to


reliably identify mode peaks in a given histogram
• The chances of selecting a “good” threshold are
enhanced if mode peaks are
– tall
– narrow
– symmetric
– and separated by deep valleys
• One approach for improving the histogram shape is
to consider only those pixels that lie on or near a
boundary between objects and the background
Electrical & Computer Engineering Dr. D. J. Jackson Lecture 18-6
Boundary characteristic thresholds
(continued)
• An obvious advantage is that the histogram
becomes less dependent on the size of objects in
the image
• By choosing pixels on or near object boundaries (
assuming an equal probability of choosing a pixel
on the object or boundary) the histogram peaks
tend to be made more symmetric
• Using pixels that satisfy some simple measures
based on the gradient and Laplacian operators
tends to deepen the valleys between histogram
peaks
Electrical & Computer Engineering Dr. D. J. Jackson Lecture 18-7

Boundary characteristic thresholds


(continued)
• Determining if a pixel lies on a boundary: compute
the gradient
• Determining what side, background (dark) or object
(light), a pixel lies on: compute the Laplacian
• Using the gradient and Laplacian, a three-level
image may be formed according to
⎧0 if ∇f < T
⎪
s ( x, y ) = ⎨+ if ∇f ≥ T and ∇ 2 f ≥ 0
⎪− if ∇f ≥ T and ∇ 2 f < 0
⎩

• where 0, + and - are three distinct gray levels


Electrical & Computer Engineering Dr. D. J. Jackson Lecture 18-8
Boundary characteristic thresholds
(continued)
• For a dark object on a light background, s(x,y) is
produced where
– all pixels not on an edge are labeled 0
– all pixels on the dark side of an edge are labeled +
– all pixels on the light side of an edge are labeled -
• For a light object on a dark background, s(x,y) is
produced where
– all pixels not on an edge are labeled 0
– all pixels on the dark side of an edge are labeled -
– all pixels on the light side of an edge are labeled +

Electrical & Computer Engineering Dr. D. J. Jackson Lecture 18-9

You might also like

pFad - Phonifier reborn

Pfad - The Proxy pFad of © 2024 Garber Painting. All rights reserved.

Note: This service is not intended for secure transactions such as banking, social media, email, or purchasing. Use at your own risk. We assume no liability whatsoever for broken pages.


Alternative Proxies:

Alternative Proxy

pFad Proxy

pFad v3 Proxy

pFad v4 Proxy