Unit 2 (Slides) PDF

DIGITAL IMAGE PROCESSING - 2
Lecture 13: Morphological

Image Processing
Dr. Shruthi M.L.J.
Department of Electronics &
Communication Engineering
1
Unit 2: Morphological Image Processing
Dr. Shruthi M.L.J.

Department of Electronics & Communication Engineering
2
Binary Morphological Operations
• Dilation
• Erosion
• Opening
• Closing
• Hit-or-Miss transform
5
Dilation and Erosion
Two fundamental morphological operations:
• Dilation
• Erosion
• Many of the morphological algorithms are based on these two

primitive operations.
6
of morphological filters in Sections 9.4 and 9.8.
Dilation
2
s sets in Z
• With A, the
and dilation
B as setsofinAZ2by B, denoted
, Dilation as A {byBstructuring
of a set , is defined as
element B:
{
A { B = z P( Bˆ ) z ¨ A ¹ Æ } (9-6)
• Reflection of B about origin and shifting this reflection by z

• Dilation of A by B is set of all displacements, z, such
that B̂ and A overlap by at least one element
• Grow the object (Does the structuring element hit the set?)
7
Dilation
• When an image X is dilated by structuring element B, the

outcome image z would be such that there will be at least one
element in B that intersects with an element in X
• If this is the case, the position where structuring element is

being centered on the image will be set to ‘1’(black).
8
Example of Dilation
• Consider Image I, composed of set (object) A and background

and square SE B (the dot is the origin)
• Perform the dilation operation using
• Dilation of A by B is
9
Example of Dilation
• Consider Image I, composed of set (object) A and background

and square SE B (the dot is the origin)
• Perform the dilation operation using
• Dilation of A by B is
10
Example of Dilation : Bridging Gaps
• Consider a low-resolution text showing broken characters

(see magnified view) ‘A’ . The maximum length of the breaks
is known to be two pixels
• And the Structuring element ‘B’
• Use white (1) to denote the foreground and black (0) for the
background
• Dilation of ‘A’ by ‘B’
• Broken segments were joined
11
Example of Dilation : Bridging Gaps
• The same could have been done using filtering

• One important advantage of the morphological
approach over the lowpass filtering method used to
bridge the gaps is that the morphological method
resulted directly in a binary image
• Lowpass filtering, on the other hand, starts with a
binary image and produces a grayscale image that
would require thresholding to convert it back to
binary form
12
Morphological Image Processing: Erosion
Dr. Shruthi M.L.J.

2
✓Dilation
• Erosion
• Opening
• Closing
5
Two fundamental morphological operations:
• Dilation
• Erosion
• Many of the morphological algorithms are based on these two

primitive operations.
6
Erosion
• With A and B as sets in Z 2 , the erosion of A by B, denoted as ,

is defined as
where A is a set of foreground pixels, B is a structuring element,

and the z’s are foreground values (1’s)
• The erosion of A by structuring element B is the set of all
points z such that B, translated by z, is contained in A
• Does the structuring element fit the set?
• Shrink the object

7
Erosion
• We work with sets of foreground pixels embedded in a set of

background pixels to form a complete image, I.
• Thus, inputs and outputs of morphological procedures are
images, not individual sets
• Hence the erosion equation can also be written as
• We use the notation when a morphological operation

uses only foreground elements, and
when the operation uses foreground and background
elements.
8
Erosion
• It is counter process of dilation

• Shrinking is determined by the structuring element
• Normally smaller than the image (3x3) to ensure faster
computation time
• Structuring element moves from left to right and top to bottom
• At centre position indicated by centre of the structuring
element, the process will look for complete overlap with image
• If not the centre pixel indicated by centre of structuring
element will be set to ‘0’(white)
9
Examples of Erosion
• Consider an Image I, consisting of a set (object) A, and

background.
• Consider square SE, B (the dot is the origin).
• Erosion of A by B
shaded region in the resulting image
10
Examples of Erosion
• Consider an Image I, consisting of a set (object) A, and

background.
• Consider elongated SE, B (the dot is the origin).
• Erosion of A by B
shaded region in the resulting image

11
Examples of Erosion
• Using erosion remove image components.

• Consider a 486 × 486 binary image of a wire-bond mask in
which foreground pixels are shown in white.
• Erode the Image using square structuring elements of sizes
11×11, 15×15, and 45 × 45 elements, respectively, all valued
1(white).
• Increasing the size of the structuring element eliminates larger

components.
12
Erosion Cont..
• Erosion shrinks or thins objects in a binary image.
• Erosion can be viewed as a morphological filtering operation

in which image details smaller than the structuring element
are filtered (removed) from the image.
13
• Dilation and Erosion are dual of each other w.r.t. set

complementation and reflection
and
• Erosion of A by B is the complement of the dilation of

Ac by and vice versa.
14
• The duality property is useful when the structuring element

values are symmetric with respect to its origin so that B =
• Then, we can obtain the erosion of A simply by dilating its
background (i.e., dilating Ac ) with the same structuring
element and complementing the result
15
Application
• Dilation
• Filling of holes of certain shape and size, given by SE
• Erosion
• Removal of structures of certain shape and size, given by SE
16
Example 1
• Eliminate irrelevant detail from binary image

• Consider binary image with squares of sizes 1,3,5,7,9 and 15
pixels
• We want to eliminate all squares except largest ones
17
Example 1
• Can be done by eroding the image with a structuring element

of size somewhat smaller than the object we wish to keep.
• Lets choose SE of size 13 x 13
• Result of eroding is (only portion of largest square remains)
18
Example 1
• The squares are shrunk

• We can restore to original size 15 x 15 by dilating them with
same structuring element used for erosion
19
Process
structuring element B = 13x13 pixels of gray level 1(white)
20
Example 2
• Using input image and structuring element as given below find

the dilated version of the input image
 Initially structuring element is aligned

with centre at *
21
Morphological Image Processing:

Dr. Shruthi M.L.J.

2
• Dilation and Erosion are dual of each other w.r.t. set

complementation and reflection
and
• Erosion of A by B is the complement of the dilation of

Ac by and vice versa.
5
• The duality property is useful when the structuring element

values are symmetric with respect to its origin so that B =
• Then, we can obtain the erosion of A simply by dilating its

background (i.e., dilating Ac ) with the same structuring
element and complementing the result
6
Application
• Dilation
• Filling of holes of certain shape and size, given by SE
• Erosion
• Removal of structures of certain shape and size, given by SE
7
Example 2

 Initially structuring element is aligned

with centre at *
8
Example 2
• The dilated image is
9
Example 1

Reflected SE
10
Example 1
• Step 1:First Row operation
• Step 2: Translate SE to right by one pixel and repeat
11
Example 1
• Step 3:Translate SE to right by one pixel and repeat
• Step 4:Translate SE to right by one pixel and repeat
12
Example 1
• Step 5: Resultant Image
Shift structuring element from top to bottom

and move from left to right again
• Step 6: Dilated Image
13
Example 3
the eroded version of the input image
14
Example 3
• Step 1: First Row operation
• Step 2: Second Row operation
15
Example 3
• Last Step: Last Row operation
Eroded Image
16
Example 4

the eroded version of the input image
17
Combining Erosion and Dilation
• Requirement:
• Remove structures / fill holes without affecting remaining
parts
• Solution:
• Combine erosion and dilation (using same SE)
• Results in two operations:
• Opening
• Closing
18
Opening
• Erosion followed by dilation, denoted by
• Eliminates protrusions
• Breaks necks
• Smoothes contour
• The opening of A by B is the union of all the
translations of B so that B fits entirely in A.
19
Opening
• Consider an image containing a set (object) A and

background
• and a solid, circular structuring element, B.
• Some of the translations of B such that it is contained
within A is as below:
• And the union of all such possible translations is set

shown shaded
20
Opening and Closing Operations
Dr. Shruthi M.L.J.

2
Combining Erosion and Dilation
• Requirement:
• Remove structures / fill holes without affecting remaining
parts
• Solution:
• Combine erosion and dilation (using same SE)
• Results in two operations:
• Opening
• Closing
5
Opening
• Erosion followed by dilation, denoted by
• Eliminates protrusions
• Breaks necks
6
Opening
• Consider an image containing a set (object) A and

background
• And a solid, circular structuring element, B.
• Some of the translations of B such that it is contained
within A (Opening) is as below:
7
Opening
• Observe that, in this case, the opening is a set composed of
two disjoint subsets, resulting from the fact that B could
not fit in the narrow segment in the center of A.
• So the opening of A by B is also given by the union of all the
translations of B so that B fits entirely in A.
• The ability to eliminate regions narrower than the
structuring element is one of the key features of
morphological opening
• The interpretation that the opening of A by B is the union
of all the translations of B such that B fits entirely within A
can be written in equation form as
8
Where ́ denotes the union of the sets inside the braces.
Opening
9
Opening
10
Closing
• Dilation followed by erosion, denoted by
• Fuse narrow breaks and long thin gulfs
• Eliminate small holes
• Fill gaps in the contour
11
Closing
• Closing has a similar geometric interpretation an opening,

except that now we translate B outside A.
• The closing is then the complement of the union of all
translations of B that do not overlap A.
12
Closing
13
Closing
• Note that the boundary of the closing is determined by the

furthest points B could reach without going inside any part of
A.
• Based on this interpretation, we can write the closing of A by B
as
14
Example 2
• Consider an image containing a single object (set) A, and a disk

structuring element
• Various positions of the structuring element during erosion
15
Example 2
• This process resulted in the disjoint set
• The bridge between the two main sections was eliminated

• Its width was thin in relation to the diameter of the structuring
element, which could not be completely contained in this part
of the set, thus violating the definition of erosion
• The same was true of the two rightmost members of the
object. Protruding elements where the disk did not fit were
eliminated. 16
Example 2
Opening
Morphological opening removes
regions that cannot contain the
structuring element, smoothens
object contours, breaks thin
connections, and removes thin
protrusions.
17
The final result of opening
Example
Closing
• As with opening, closing also
smoothens the contours of objects.
• However, unlike opening, closing
tends to join narrow breaks, fills
long thin gulfs, and fills objects
smaller than the structuring
element.
• In this example, the principal result
of closing was that it filled the
small gulf on the left of set A.
18
Use of Open and Close
19
Properties
• Opening
1. A°B is a subset (subimage) of A
2. If C is a subset of D, then C °B is a subset of D °B
3. (A °B) °B = A °B
• Closing
1. A is a subset (subimage) of A•B
2. If C is a subset of D, then C •B is a subset of D •B
3. (A •B) •B = A •B
• Note: repeated openings/closings has no effect!
20
Duality
• Dilation and Erosion are dual with respect to

set complementation and reflection • Then, we can obtain the erosion of A simply
by dilating its background (i.e., dilating Ac )
with the same structuring element and
complementing the result
• Opening and closing are also dual with respect

to set complementation and reflection
21
✓Dilation: Grow the object

:Does SE hit the set?
✓Erosion: Shrink the object
✓ : :Does SE fit the set?
✓Opening: Breaks necks
✓Closing: Fill gaps/Fuse narrow breaks
22
Hit or Miss Transformation ⊛ (HMT)
• Basic tool for shape detection

• Find location of one shape among a set of shapes “template
matching”
• Utilizes two Structuring Elements:

• B1: for detecting object part/ shapes in foreground and
• B2 : for detecting shapes in background
• Does B1 fit the object while, simultaneously, B2 misses the

object, i.e., fits the background? 23
Hit or Miss Transformations (HMT)
Dr. Shruthi M.L.J.

2
✓Dilation: Grow the object

:Does SE hit the set?
✓Erosion: Shrink the object
✓ : :Does SE fit the set?
✓Opening: Breaks necks
✓Closing: Fill gaps/Fuse narrow breaks
4
Hit or Miss Transformation ⊛ (HMT)
• Basic tool for shape detection

• Find location of one shape among a set of shapes “template
matching”
• Utilizes two Structuring Elements:

• B1: for detecting object part/ shapes in foreground and
• B2 : for detecting shapes in background
• Does B1 fit the object while, simultaneously, B2 misses the

object, i.e., fits the background? 5
HMT Cont…
• Let I be a binary image composed of foreground (A) and

background pixels, respectively
• The HMT of image I is defined as
; (Follows from definition of erosion)
• The morphological HMT is the set of translations, z, of

structuring elements B1 and B2 such that, simultaneously, B1
found a match in the foreground (i.e., B1 is contained in A) and
B2 found a match in the background (i.e., B2 is contained in Ac )
6
HMT Cont…
• The word “simultaneous” implies that z is the same translation of

both structuring elements.
• The word “miss” in the HMT arises from the fact that B2 finding a
match in Ac is the same as B2 not finding (missing) a match in A.
7
Example 1
• Find the hit or miss transformation(HMT) of the input image

whose Structuring Elements B1 and B2 are as given below
8
Example 1
• Step 1: Find erosion of input image with B1
• Find erosion of complement of input image with B2
9
Example 1
• Step 3: Find intersection of results of steps 1 and 2
10
HMT
• If we use foreground and background pixels in one SE we

need not use 2 SEs.
• Then HMT is then defined as
• Similar to erosion definition with difference that we now

check if the (B)z is subset of image I, which is composed
of both foreground and background pixels
• This formulation is general and B can be structured to
detect any arrangement of pixels in image I
11
Example 2
Examples of using a single structuring element

using following equation to detect specific
features
• Ex 2.1 : detection B of single-pixel holes
12
Example 2
Ex 2.2 : detection of an upper-right corner
Ex 2.3 : detection of multiple features
13
Example 2
• The example 2.3 has structuring element composed of foreground,

background, and “don’t care” elements denoted by ×’s
• The value of a don’t care elements can be thought of as always matching
its corresponding pixel in an image.
• In this example, when the SE is centered on the top, right corner pixel, the
don’t care elements in the top of the SE can be considered to be
background, and the don’t care elements on the bottom row as
foreground, producing a correct match
• When the SE is centered on the bottom, right corner pixel, the role of the
don’t care elements is reversed, again resulting in a correct match.
• The other border pixels between the two corners were similarly detected
by considering all don’t care elements as foreground.
• Thus, using don’t care elements increases the flexibility of structuring
elements to perform multiple roles. 14
Example 3
• Find the location of the origin of object (set) D in image I.
Image consisting of a foreground

(1’s) equal to the union, A, of set of
objects, and a background of 0’s.
• Here, A is the union of all object sets, so D is a subset of A.

15
Example 3
• All three objects are composed of foreground pixels

• One way of explaining why they appear as different shapes is
because each occupies a different area of the background.
• In other words, the nature of a shape is determined by the
geometrical arrangement of both foreground and background
pixels.
Image with
its foreground defined as Ac . Structuring elements designed to detect object D. 16
Element B1 is equal to D itself
Next Session
• HMT Cont…
• Morphological Algorithms
17
THANK YOU
Dr. Shruthi M.L.J.

shruthimlj@pes.edu
+91 9482219115
18
DIGITAL IMAGE PROCESSING – 2
Lecture 18: Morphological

Image Processing
Dr. Shruthi M.L.J.
1
Morphological Algorithms
Dr. Shruthi M.L.J.

2
Last Session
• Hit or Miss Transform
3
This Session
• Hit or Miss Transform Cont..

4
HMT Cont…
• Let I be a binary image composed of foreground (A) and

background pixels, respectively
• The HMT of image I is defined as
; (Follows from definition of erosion)
• The morphological HMT is the set of translations, z, of

structuring elements B1 and B2 such that, simultaneously, B1
found a match in the foreground (i.e., B1 is contained in A) and
B2 found a match in the background (i.e., B2 is contained in Ac )
5
Example 3
• Find the location of the origin of object (set) D in image I.
Image consisting of a foreground

(1’s) equal to the union, A, of set of
objects, and a background of 0’s.
• Here, A is the union of all object sets, so D is a subset of A.

6
Example 3
• All three objects are composed of foreground pixels
Image I with A as foreground Image with its foreground defined as Ac

7
Example 3
• We need two structuring elements

• Structuring element B1 is equal to D itself
• Structuring element B2 is designed to detect D in Ic .

• Because D is composed of background elements in Ic, and
erosion works with foreground elements, B2 has to be
designed to detect the border of D, which is composed of
foreground pixels in Ic
• It consists of a rectangle of foreground elements one pixel
thick. The size of the rectangle is such that it encloses the size
of D
8
Example 3
Structuring elements designed

Image with its foreground
to detect object D.
defined as Ac . Image with its foreground defined as Ac
Erosion of A by B1 contains a single

point: the origin of D, as desired, but
it also contains parts of object C
9
Erosion of A by B1.
Example 3
*Erosion works with foreground elements
Structuring element
designed to detect object D
Image with its foreground defined as Ac
• The erosion of the foreground of Ic by

B2 is the shaded region .
• It contains the origin of D, but is also
contains parts of sets Ac and C
Erosion of Ac by B2 10
Example 3
Image with its foreground Structuring elements designed

to detect object D. Image with its foreground defined as Ac
defined as A
Shows the location of the origin

of D, as desired.
Erosion of A by B1. Intersection of the 2 erosions
Using single SE
• Why not try to detect D directly in image I using a single

structuring element, instead of going thorough such a
laborious process
• In order to detect D directly in image I, we should be able to
process foreground and background pixels simultaneously,
rather than processing just foreground pixels, as required by
the definition of erosion.
• We achieve this by defining a structuring element, B, identical
to D, but having in addition a border of background elements
with a width of one pixel.
12
Using single SE
• We can use the structuring element formed in such a way to
restate the HMT as
• Here we test to see if (B)z is a subset of image I, which is

composed of both foreground and background pixels. Image with its foreground
• For example 3, Design a single structuring element to detect defined as A
D(border of D)
13
Example 3 (using single SE)
• So using a single SE
14
Some Basic Morphological Algorithms
• After basic morphological operations, we are now ready to

consider some practical uses of morphology.
• One of the principal applications of morphology (dealing with
binary images) is in extracting image components that are useful in
the representation and description of shape
• In particular, we consider morphological algorithms for extracting
boundaries, connected components, the convex hull, and the
skeleton of a region.
• We also develop several methods (for region filling, thinning,
thickening, and pruning) that are used frequently for pre- or post-
processing.
15
• Boundary Extraction
• Region Filling
• Extraction of connected components
• Convex hull
• Thinning
• Thickening
• Skeletons
• Pruning
16
Morphological Algorithms
Dr. Shruthi M.L.J.

2
Example 3
*Erosion works with foreground elements
Structuring element
designed to detect object D
Image with its foreground defined as Ac
• The erosion of the foreground of Ic by

B2 is the shaded region .
• It contains the origin of D, but is also
contains parts of sets Ac and C
Example 3
Image with its foreground Structuring elements designed

to detect object D. Image with its foreground defined as Ac
defined as A
Shows the location of the origin

of D, as desired.
Erosion of A by B1. Intersection of the 2 erosions
• After basic morphological operations, we are now ready to

consider some practical uses of morphology.
• One of the principal applications of morphology (dealing with
binary images) is in extracting image components that are useful in
the representation and description of shape
• In particular, we consider morphological algorithms for extracting
boundaries, connected components, the convex hull, and the
skeleton of a region.
• We also develop several methods (for region filling, thinning,
thickening, and pruning) that are used frequently for pre- or post-
processing.
6
• Region Filling
• Convex hull
• Thinning
• Thickening
• Skeletons
• Pruning
7
Boundary Extraction
• The boundary of set A of foreground pixels is obtained by first

eroding A by a suitable SE B, and then performing the set
difference between A and its erosion
8
Boundary Extraction
Example 1: Consider a simple binary object and a structuring

element B
Structuring
Set, A, of foreground pixels. element. A eroded by B.
Boundary of A.
9
Boundary Extraction
Example 2: Consider a simple binary object and a structuring

element B
Structuring
element.
A binary image
Boundary of A.
10
• Region Filling
• Convex hull
• Thinning
• Thickening
• Skeletons
• Pruning
11
Region / Hole Filling
• Hole: defined as background region surrounded by connected border

of foreground pixels
• Region/hole filling: The process of ‘coloring in’ a definite image area

or region
• Algorithm is based on set dilation, complementation and intersection

for filling holes/regions in an image
12
n image. Let A denote a set whose elements are 8-connected bound-
The Algorithm
boundary enclosing a background region (i.e., a hole). Given a point
objective
• Step 1:is to fill all the holes with foreground elements (1’ s).
orming
• Formananarray, X0 0of, 0’s
array , X 0’s (the
of (same size same
as I, thesize ascontaining
image I , the image
A) containing
ationsexcept
in Xat0locations in X0 that correspond
that correspond to pixels to pixels
thatthat
areareknown to be holes,
known to be holes, which we set to 1. Then following
1. Then, the following procedure fills all the holes with 1’ s:
procedure fills all the holes with 1’s
X k = ( X k -1 { B ) ¨ I c
k = 1, 2, 3, … (9-19)
• B is symmetric
ymmetric SE
structuring element in Fig. 9.17(c) . The algorithm termi-
• The
n step k if X k = terminates
algorithm at iteration
X k -1. Then, step k if Xall
X k contains k=X the
k-1 filled holes. The set
• Union ofall
I contains original image Iholes
the filled with Xkand
results in region
their filling
boundaries.
n Eq. (9-19) would fill the entire area if left unchecked, but the
ach step with I c limits the result to inside the region of interest. This is 13
DIGITAL IMAGEfilling holes in an image. Let A denote a set whose elements are 8-connected bound-
PROCESSING -2
aries, with each boundary enclosing a background region (i.e., a hole). Given a point
Example 1 in each hole, the objective is to fill all the holes with foreground elements (1’ s).
We begin by forming an array, X 0 , of 0’s (the same size as I , the image containing
A), except at locations in X 0 that correspond to pixels that are known to be holes,
• The input image
whichand
we structuring the following
element
set to 1. Then, procedure
are shown fills all the holes with 1’ s:
below.
Remember, the dila-
Perform region filling operation
X k = ( X k -1 { B ) ¨ I c
tion of image X by B
is the dilation of the k = 1, 2, 3, … (9-19)
foreground
elements of X by B.
where B is the symmetric structuring element in Fig. 9.17(c) . The algorithm termi-
nates at iteration step k if X k = X k -1. Then, X k contains all the filled holes. The set
union of X k and I contains all the filled holes and their boundaries.
The dilation in Eq. (9-19) would fill the entire area if left unchecked, but the
intersection at each step with I c limits the result to inside the region of interest. This is
our first example of how a morphological process can be conditioned to meet a desired
property. In the current application, the process is appropriately called conditional
Input Image Structuring
dilation. The restElement
of Fig. 9.17 illustrates furtherof
Complement the mechanics
Input Imageof Eq. (9-19). This exam-
ple only has one hole, but the concept applies to any finite number of holes, assuming
that a point inside each hole is given (we remove this requirement in Section 9.6).
14
E
XAMPE9.6: M
L orph
ologic
al h
olefillin
g.
by forming an array, X 0 , of 0’s (the same size as I , the image containing
DIGITAL IMAGE
locations in X 0PROCESSING -2
that correspond to pixels that are known to be holes,
Then,1the
Example
to 1. following procedure fills all the holes with 1’ s:
: Solution
X k = ( X k -1 { B ) ¨ I c
k = 1, 2, 3, … (9-19)
• Step 1: Initially take X0 as shown below.
e symmetric
• Perform dilationstructuring
of X with SE B.
0
element in Fig. 9.17(c) . The algorithm termi-
tion •• step k if X k = X k -1. Then, X k contains all the filled holes. The set
The resultant image is then intersected with complement of input image, I.
This completes first iteration.
and I contains all the filled holes and their boundaries.
on in Eq. (9-19) would fill the entire area if left unchecked, but the
t each step with I c limits the result to inside the region of interest. This is
ple of how a morphological process can be conditioned to meet a desired
he current application, the process is appropriately called conditional
est of Fig. 9.17 illustrates further the mechanics of Eq. (9-19). This exam-
one hole, but the concept applies to any finite number of holes, assuming
15
nside each hole is given (we remove this requirement in Section 9.6).
Example 1 : Solution
• Step 2: Repeat Step 1 with input as output of step 1
• Step 3: Repeat same process till no change
16
• Step 4 and Step 5:
17
• Step 6:
• Step 7: Perform union of result of Step 6 and original image to

get region filled image
18
by forming an array, X 0 , of 0’s (the same size as I , the image containing
locations in X 0 that correspond to pixels that are known to be holes,
Example
to 1. Then,2the following procedure fills all the holes with 1’ s:
X k = ( X k -1 { B ) ¨ I c k = 1, 2, 3, … (9-19)
he symmetric structuring element in Fig. 9.17(c) . The algorithm termi-

ation step k if X k = X k -1. Then, X k contains all the filled holes. The set
and I contains all the filled holes and their boundaries.
on in Eq. (9-19) would fill the entire area if left unchecked, but the
at each step with I c limits the result to inside the region of interest. This is
mple of how a morphological process can be conditioned to meet a desired
the current application, the process is appropriately called conditional
rest of Fig. 9.17 illustrates further the mechanics of Eq. (9-19). This exam-
one hole, but the concept applies to any finite number of holes, assuming
nside each hole is given (we remove this requirement in Section 9.6).
19
✓Boundary Extraction
✓Region Filling
• Convex hull
• Thinning
• Thickening
• Skeletons
• Pruning
20
Morphological Algorithms Cont..
Dr. Shruthi M.L.J.

2
✓Boundary Extraction
✓Region Filling
• Convex hull
• Thinning
• Thickening
• Skeletons
• Pruning
4
Extraction of Connected Componenets
• Extraction of connected components from binary image is

required in many automated image analysis applications
• Step 1: Initially take X0 (of same size as I, the image containing
A) whose elements are 0’s, except at each location known to
correspond to a point in each connected component in A,
which is to be set to 1
• Objective is to start with X0 and find all connected components
in I
5
Extraction of Connected Components
• The procedure terminates when Xk = Xk-1
• Xk contains all connected components of foreground pixels in

the image
• Here we intersect with I instead of Ic (as in region filling), as we

are looking for foreground points, while in hole filling objective
is to find background points
6
Example
7
• Region Filling
• Convex hull
• Thinning
• Thickening
• Skeletons
• Pruning
8
Convex Hull
• A set ‘A’ is said to be convex if the straight line segment

joining any two points in ‘A’ lie entirely within A.
• Convex hull, H, is the smallest convex set containing A
• Let denote 4 structuring elements as below
9
Convex Hull
• The procedure consists of implementing the morphological

equation
with
When procedure converges using ith SE (when we let
• Then convex hull of A is union of all 4 results
10
Convex Hull
• This method consists of iteratively applying hit or miss

transform to I with Bi until convergence, then letting Di=Xki,
where k is the step where convergence occurred
• Procedure is repeated with B2 (applied to I) until no further

changes occur and so on
• Union of 4 resulting Di constitutes the convex hull of A
• Initialize with k=0 and X0i = I every time SE changes
11
Example 1
Determine the Convex hull, H, of the given image I
Structuring elements.
Convex hull, H, is the smallest convex set containing A
12
Example 1
13
Example 1
14
Example 1
15
Example 1
16
Example 1
17
Example 1
18
• Region Filling
• Convex hull
• Thinning
• Thickening
• Skeletons
• Pruning
19
Dr. Shruthi M.L.J.

2
• Region Filling
• Convex hull Cont..
• Thinning
• Thickening
• Skeletons
• Pruning
4
Convex Hull
• A set, S, of points in the Euclidean plane is said to be convex if and

only if a straight line segment joining any two points in S lies
entirely within S.
• The convex hull, H, of S is the smallest convex set containing S.
Convex Set
Not a Convex Set

5
Convex Hull
• The digital image plane only allows points at discrete

coordinates.
• Thus, the sets with which we work are digital sets.
• The same concepts of convexity are applicable to digital sets,
but the definition of a convex digital set is slightly different.
• A digital set, A, is said to be convex if and only if its Euclidean
convex hull only contains digital points belonging to A.
• A simple way to visualize if a digital set of foreground points is
convex is to join its boundary points by straight (continuous)
Euclidean line segments.
• If only foreground points are contained within the set formed
by the line segments, then the set is convex; otherwise it is
not.
6
Algorithm to determine Convex Hull
• Obtain an approximation of the convex hull of a set

A of foreground pixels, embedded in a binary image, I.
• Let denote 4 structuring elements as below
7
Algorithm to determine Convex Hull
• The procedure consists of implementing the morphological

equation
with
When procedure converges using ith SE
• Then convex hull of A is union of all 4 results
8
Convex Hull
• This method consists of iteratively applying hit or miss

transform to I with Bi until convergence, then letting Di=Xki,
where k is the step where convergence occurred
• Procedure is repeated with B2 (applied to I) until no further

changes occur and so on
• Union of 4 resulting Di constitutes the convex hull of A
• Initialize with k=0 and X0i = I every time SE changes
9
Example
Determine the Convex hull, H, of the given image I
Convex hull, H, is the smallest convex set containing A
10
Example: First ISE, i=1
Structuring element
X1 1 X1 2
11
X1 X1 4
Example: Second SE, i=2
Structuring element
X2 0 X2 1 X2 2
12
Example: Third SE, i=3
Structuring element
13
Example: Fourth SE, i=4
Structuring element
X4 1
14
Example 1
15
Example 1
*This method does

not yield the smallest
possible Hull
16
• Region Filling
• Convex hull
• Thinning
• Thickening
• Skeletons
• Pruning
17
Thinning
• The thinning of a set A by a structuring element B, is defined as
• A more useful expression for thinning A symmetrically is based

on a sequence of structuring elements:
• where Bi is rotated version of Bi-2

• The thinning by a sequence of structuring element {B}
18
Example
• The process is to thin A by one pass with B1
• Then thin the result with one pass of B2, and so on, until A is
thinned with one pass of Bn.
• The entire process is repeated until no further changes occur

after one complete pass through all structuring elements.
• Each individual thinning pass is performed using the given

equation
19
Example
• Consider a set of structuring elements used routinely for thinning
(Bi is equal to Bi−1 rotated clockwise by 45° )
20
Example
21
Example
22
Example
23
• Region Filling
• Convex hull
• Thinning
• Thickening
• Skeletons
• Pruning
24
ADVANCED DIGITAL IMAGE PROCESSING
Dr. Shruthi M.L.J.

2
• Region Filling
• Convex hull
• Thinning
• Thickening
• Skeletons
• Pruning
5
Thickening
• Morphological operation used to grow selected regions of
foreground pixels
• Thickening is the morphological dual of thinning and is defined
by the expression
• As in thinning, thickening can be defined as a sequential

operation:
6
Thickening
• The structuring elements used for thickening have the same form as
those shown in thinning but with all 1’s and 0’s interchanged
• However, a separate algorithm for thickening is seldom used in practice.
• Instead, the usual procedure is to thin the background of the set in

question, then complement the result.
• In other words, to thicken a set A we form Ac , thin Ac , and then

complement the thinned set to obtain the thickening of A.
• Remove all disconnected components

7
Example of Thickening using Thinning
Perform
Using
(a) Set A. (b) Complement of A (c) Thinning of Ac by B1

Ac B1
8
Example of Thickening using Thinning
Perform
Using
(a) Set A. (b) Complement of A (c) Result of thinning the complement.
(d) Thickened set obtained by 9

(e) Final result, with no disconnected points.
Analysis
• Depending on the structure of A, this procedure can result in

disconnected points
• Hence thickening by this method usually is followed by post-
processing to remove disconnected points.
• Note from (c) that the thinned background forms a boundary
for the thickening process.
• This useful feature is not present in the direct
implementation of thickening using the Eq., and it is one of
the principal reasons for using background thinning to
accomplish thickening.
10
• Region Filling
• Convex hull
• Thinning
• Thickening
• Skeletons
• Pruning
11
Skeletons
• Consider the figures below:
• Condition: If z is a point in S(A), and (D)z is the largest disk

centered at z and contained in A, one can’t find a larger disk
containing (D)z and simultaneously included in A.
• A disk satisfying these conditions is called a maximum disk
12
Skeletons
• If (D)z is a maximum disk, it touches the boundary of A at two

or more different places
• The skeleton of A can be expressed in terms of erosions and
openings
With where
indicates k successive erosions starting with ‘A’ k times
K is the last iterative step before A erodes to an empty set.

13
Example
Consider Set A. Determine its skeleton
denotes k successive dilations, starting with Sk (A)
14
Example
Consider Set A.
Various positions of maximum disks Another maximum disk, whose center

whose centers partially define the defines a different segment of the
skeleton of A. skeleton of A.
Complete skeleton (dashed).
15
• Region Filling
• Convex hull
• Thinning
• Thickening
• Skeletons
• Pruning
16
Pruning
• Complement to thinning and skeletonizing algorithms
• Generally required in post processing to clean up
• Consider automated recognition of hand written characters. Here

shape of skeleton of character is analyzed
• These skeletons often contain spurs, due to noise and non

uniformity in character strokes
• Remove these using morphological technique with the assumption

that the length of a parasitic component does not exceed a specified
number of pixels.
17
Pruning
• Consider the alphabet ‘a’. It has spurs that need to be removed
• Assumption is any branch with 3 or less pixels is to be

eliminated
• Thinning of set A with a sequence of SE designed to detect
only end points achieves the desired result
where {B} denotes the structuring element sequence

18
Pruning
• The sequence of structuring elements consists of two

different structures, each of which is rotated 90° for a
total of eight elements.
• Each SE is a detector for an end point in a particular

orientation
19
Example
• Perform pruning on the following image
Step 1: Perform thinning using the given SEs using
SEs used for deleting end points.
(c) X1 : Result of three cycles

of thinning.
20
Example
Step 2: The next step is to “restore” the character to its original form, but
with the parasitic branches removed.
This requires that we first form a set X2 containing all end points in X1
where the Bk are the end-point detectors
SEs used for deleting end points.
(d) X2 : End points of (c).
21
Example
Step 3: The next step is dilation of the end points.

Typically, the number of dilations is less than the number of end- point
removals to reduce the probability of “growing” back some of the spurs.
Here we dilate the end points three times using A as a delimiter.

This is the same number of thinning passes:
where H is a 3 X 3 structuring element of 1’s, and the intersection with A

is applied after each step
As in the case of region filling, this type of conditional

dilation prevents the creation of 1-valued elements outside
the region of interest,
X3 : Dilation of end points conditioned on (a)

22
Example
Step 4:Finally, take the union of X1 and X3 ,
X1 : Result of three cycles (f) Pruned image

X3 : Dilation of end points
of thinning.
conditioned on (a)
23
Example
(a) Set A of foreground

(c) Result of three cycles
pixels (shaded) (d) End points of (c).
of thinning.
SEs used for detecting end points.
(f) Pruned image

(e) Dilation of end points
conditioned on (a) 24
Steps involved in Pruning
• Steps for pruning
X 1 = A Ä{B} ;Thinning
where the Bk are the end-point detectors
25
Summary of
binary
morpho
logical
operations
and
their
properties.
26
Grayscale Morphology
Dr. Shruthi M.L.J.

2
Morphological Operations on Grayscale Images
• Basic operations of dilation, erosion, opening, and closing of

binary images can be extended to grayscale images
• These operations can be used to develop several basic
grayscale morphological algorithms
• Consider f(x,y) and b(x,y), where f(x,y) is a grayscale image and
b(x, y) is a structuring element
• The assumption is that these functions are discrete
• That is, if Z denotes the set of real integers, then the coordinates (x, y)
are integers from the Cartesian product Z2, and f(x,y) and b(x,y) are
functions that assign an intensity value (a real number from the set of
real numbers, R) to each distinct pair of coordinates (x, y). If the
intensity levels are integers also, then Z replaces R.
5
Structuring Elements
• Structuring elements in grayscale morphology perform the

same basic functions as their binary counterparts
• They are used as “probes” to examine a given image for
specific properties.
• Structuring elements in grayscale morphology belong to one of
two categories: nonflat and flat.
Here 0(0): Black and 1(255): White
Horizontal intensity profiles 6

through their centers
Structuring Elements
• As in the binary case, the origin of grayscale structuring

elements must be clearly identified
• Unless mentioned otherwise, all the examples in this section

are based on symmetrical, flat structuring elements of unit
height whose origins are at the center.
• The reflection of an SE in grayscale morphology is defined as
7
Grayscale Morphological Operations
• Erosion
• Dilation
• Opening
• Closing
8
Erosion
• The grayscale erosion of f by a flat structuring element b at
location (x, y) is defined as the minimum value of the image in
the region coincident with b(x, y) when the origin of b is at (x, y)
• The erosion at (x, y) of an image f by a structuring element b is
given as
where, in a manner similar to spatial correlation, x and y are

incremented through all values required so that the origin of b
visits every pixel in f.
• That is, to find the erosion of f by b, we place the origin of the
structuring element at every pixel location in the image. The
erosion at any location is determined by selecting the minimum
value of f in the region coincident with b.
9
Erosion
• For example, if b is a square structuring element of size 3 × 3,
obtaining the erosion at a point requires finding the minimum of
the nine values of f contained in the 3 × 3 region spanned by b
when its origin is at that point.
• Because grayscale erosion with a flat SE computes the minimum
intensity value of f in every neighborhood of (x, y) coincident
with b, we expect in general
• that an eroded grayscale image will be darker than the
original
• that the sizes (with respect to the size of the SE) of bright
features will be reduced
• and that the sizes of dark features will be increased
10
Erosion Example
Gray-scale X-ray image of Erosion using a flat disk SE

size 448 x 425 pixels. with a radius of 2 pixels
• The intensities of the small bright dots were reduced, making

them barely visible, while the dark features grew in thickness.
• The general background of the eroded image is slightly darker
than the background of the original image
11
Erosion with Nonflat SE

• Nonflat SEs have grayscale values that vary over their domain of
definition. The erosion of image f by nonflat structuring element,
bN , is defined as
• Here, we subtract values from f to determine the erosion at any

point
• Unlike flat SE, erosion using a nonflat SE is not bounded in
general by the values of f, which can be problematic in
interpreting results.
• Grayscale SEs are seldom used in practice because of this, the
potential difficulties in selecting meaningful elements for bN ,
and the added computational burden
12
Grayscale Dilation
• Grayscale dilation of f by a flat structuring element b at any
location (x,y) is defined as the maximum value of the image in
the window spanned by when the origin of is at (x, y)
Where
• Use the maximum, rather than the minimum operation, and

keep in mind that the structuring element is reflected about its
origin, taken into account by using (−s, −t) in the argument of the
function.
• This is analogous to spatial convolution
13
Grayscale Dilation
Gray-scale X-ray image of Erosion using a flat disk SE Dilation using the same SE.
size 448 x 425 pixels. with a radius of 2 pixels
• The effects are the opposite of using erosion. The bright features were
thickened and the intensities of the darker features were reduced.
• In particular, the thin black connecting wires in the left, middle, and right
bottom are barely visible
• The sizes of the dark dots were reduced as a result of dilation, but, unlike the
eroded small white dots, they still are easily visible in the dilated image.
• The reason is that the black dots were originally larger than the white dots with
respect to the size of the SE. Finally, observe that the background of the dilated 14
image is slightly lighter than that of (a).
Dilation with Non flat SE
• The dilation of image f by nonflat structuring element, bN , is

defined as
• This also has same issues as erosion
15
Grayscale Erosion and Dilation

• As in the binary case, grayscale erosion and dilation are duals
with respect complementation and reflection
• The same expression holds for nonflat structuring elements

• The notation can be simplified by suppressing the arguments of
all functions, in which case the preceding equation is written as
16
Grayscale Erosion and Dilation
• Erosion and dilation by themselves are not particularly useful in

grayscale image processing.
• As with their binary counterparts, these operations become

powerful when used in combination to derive higher-level
algorithms.
17
Grayscale Opening and Closing

• The expressions for opening and closing grayscale images have the
same form as their binary counterparts.
• The grayscale opening of image f by structuring element b, denoted as
is
• Opening is simply the erosion of f by b, followed by a dilation of the

result by b.
• Similarly, the grayscale closing of f by b, denoted as is
18
• The opening and closing for grayscale images are duals with
respect to complementation and SE reflection:
• Because f c = −f, we can write
19

• Opening and closing of grayscale images have a simple geometric
interpretation.
• Suppose that an image function f ( x, y) is viewed as a 3-D surface; that
is, its intensity values are interpreted as height values over the xy-plane
• Then the opening of f by b can be interpreted geometrically as pushing
the structuring element up from below against the undersurface of f.
• At each location of the origin of b, the opening is the highest value
reached by any part of b as it pushes up against the undersurface of f.
• The complete opening is then the set of all such values obtained by the
origin of b visiting every (x, y) coordinate of f.
20
Grayscale Opening with flat SE

• Consider the curve in figure below is the intensity profile along a single row of
an image.
• Now consider a flat structuring element in several positions, pushed up

against the bottom of the curve.
• Complete opening is shown by heavy curve

• Because the structuring element is too large to fit completely inside the
upward peaks of the curve, the tops of the peaks are clipped by the opening,
with the amount removed being proportional to how far the structuring
element was able to reach into the peak.
• In general, openings are used to remove small, bright details, while leaving the
overall intensity levels and larger bright features relatively undisturbed.
21
Grayscale Closing with flat SE

• Consider the curve in figure below is the intensity profile along a single
row of an image.
• Here the structuring element is pushed down on top of the curve

while being translated to all locations.
• The closing, is constructed by finding the lowest points reached by any

part of the structuring element as it slides against the upper side of the
curve
• The properties of grayscale opening and closing remain same as their

binary counterparts
22
Summary of Unit 2
• Basic Morphological operations
• Erosion
• Dilation
• Opening
• Closing
• Region Filling
• Convex hull
• Thinning
• Thickening
• Skeletons
• Pruning
• Grayscale Morphology
• Grayscale morphological operations
23

Unit 2 (Slides) PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Unit 2 (Slides) PDF

Uploaded by

Copyright:

Available Formats

DIGITAL IMAGE PROCESSING - 2

Lecture 13: Morphological

Unit 2: Morphological Image Processing

Dr. Shruthi M.L.J.

Two fundamental morphological operations:

• Many of the morphological algorithms are based on these two

• Reflection of B about origin and shifting this reflection by z

• When an image X is dilated by structuring element B, the

• If this is the case, the position where structuring element is

• Consider Image I, composed of set (object) A and background

• Perform the dilation operation using

• Consider Image I, composed of set (object) A and background

• Perform the dilation operation using

• Consider a low-resolution text showing broken characters

• Broken segments were joined

• The same could have been done using filtering

Morphological Image Processing: Erosion

Dr. Shruthi M.L.J.

Two fundamental morphological operations:

• Many of the morphological algorithms are based on these two

• With A and B as sets in Z 2 , the erosion of A by B, denoted as ,

where A is a set of foreground pixels, B is a structuring element,

• Does the structuring element fit the set?

• Shrink the object

• We work with sets of foreground pixels embedded in a set of

• We use the notation when a morphological operation

• It is counter process of dilation

• Consider an Image I, consisting of a set (object) A, and

shaded region in the resulting image

• Consider an Image I, consisting of a set (object) A, and

shaded region in the resulting image

• Using erosion remove image components.

• Increasing the size of the structuring element eliminates larger

• Erosion shrinks or thins objects in a binary image.

• Erosion can be viewed as a morphological filtering operation

• Dilation and Erosion are dual of each other w.r.t. set

• Erosion of A by B is the complement of the dilation of

• The duality property is useful when the structuring element

• Eliminate irrelevant detail from binary image

• We want to eliminate all squares except largest ones

• Can be done by eroding the image with a structuring element

• The squares are shrunk

structuring element B = 13x13 pixels of gray level 1(white)

• Using input image and structuring element as given below find

 Initially structuring element is aligned

Morphological Image Processing:

Dr. Shruthi M.L.J.

• Dilation and Erosion are dual of each other w.r.t. set

• Erosion of A by B is the complement of the dilation of

• The duality property is useful when the structuring element

• Then, we can obtain the erosion of A simply by dilating its

• Using input image and structuring element as given below find

 Initially structuring element is aligned

• The dilated image is

• Using input image and structuring element as given below find

• Step 1:First Row operation

• Step 2: Translate SE to right by one pixel and repeat

• Step 3:Translate SE to right by one pixel and repeat

• Step 4:Translate SE to right by one pixel and repeat

• Step 5: Resultant Image

Shift structuring element from top to bottom

• Step 1: First Row operation

• Step 2: Second Row operation