Chapter10 Segmentation PDF

Image Segmentation
Element of Image Analysis
Preprocess
Image acquisition, restoration, and enhancement
Intermediate process
Image segmentation and feature extraction
High level process

Image interpretation and recognition
ISAN-DSP GROUP
Importance of Image Segmentation
Image segmentation is used to separate an image into constituent
parts based on some image attributes. Image segmentation is an
important step in image analysis
Benefit
1. Image segmentation reduces huge amount of unnecessary
data while retaining only importance data for image analysis
2. Image segmentation converts bitmap data into better
structured data which is easier to be interpreted
Image Attributes for Image Segmentation
1. Similarity properties of pixels inside the object are used to group
pixels into the same set.
2. Discontinuity of pixel properties at the boundary between object
and background is used to distinguish between pixels belonging to
the object and those of background.
Discontinuity:
Intensity change
at boundary
Similarity:
Internal
pixels share
the same
intensity
Spatial Filtering Application to Shape Detection
 One application of spatial filtering is shape detection: finding

locations of objects with the desired shape.
 Unlike frequency selective masks that are designed based

on the concept of frequency, shape detection masks are
derived from the shapes to be detected themselves.
 A mask for shape detection usually contains the shape or a part

of the shape to be detected.
 The location that is most correlated to the mask is the

location where the highest filter response occurs. The
shape is most likely to exist there.
Point Detection
 We can use Laplacian masks
-1 -1 -1 0 -1 0
for point detection.
-1 8 -1 -1 4 -1
 Laplacian masks have the largest -1 -1 -1 0 -1 0
coefficient at the center of the mask
while neighbor pixels have an
opposite sign.
 This mask will give the high response to the object that has the
similar shape as the mask such as isolated points.
 Notice that sum of all coefficients of the mask is equal to zero.

This is due to the need that the response of the filter must be zero
inside a constant intensity area
Point Detection
Point detection can be done by applying the thresholding function:
1 f ( x, y )  T
g ( x, y )  
0 otherwise
Location of
porosity
X-ray image of the Laplacian image After thresholding
turbine blade with
porosity (Images from Rafael C. Gonzalez and Richard E.
Wood, Digital Image Processing, 2nd Edition.
Line Detection
 Similar to point detection, line detection can be performed

using the mask the has the shape look similar to a part of a line
 There are several directions that the line in a digital image can be.
 For a simple line detection, 4 directions that are mostly used are
Horizontal, +45 degree, vertical and –45 degree.
Line detection masks

(Images from Rafael C. Gonzalez and Richard E.
Line Detection Example
Binary wire
bond mask
image
Absolute value
of result after Result after
processing with thresholding
-45 line detector
Notice that –45 degree

lines are most sensitive
(Images from Rafael C.

Gonzalez and Richard E.
Wood, Digital Image
Processing, 2nd Edition.
Edges
Ideal step edge Ideal ramp edge
Blurred edge
Generally, objects and background have different intensities. Therefore,

Edges of the objects are the areas where abrupt intensity changes occur.

Ideal Ramp Edges and its Derivatives
Edge
Original image

Smoothed Step Edge and Its Derivatives
1.2
Intensity
1 Edge Edge
0.8
Gray level profile 0.6
0.4
0.2
0
-0.2
-1.5 -1 -0.5 0 0.5 1 1.5 2 2.5 3 3.5
0.06 Minimum
0.04
0.02 point
The 1st derivative 0
-0.02
-0.04 Maximum
-0.5point
-0.06
-1.5 -1 0 0.5 1 1.5 2 2.5 3 3.5
x 10 -3 Zero
5
4
3
crossing
2
1
+ +
The 2nd derivative 0
-1 - -
-2
-3
-4
-5
-1.5 -1 -0.5 0 0.5 1 1.5 2 2.5 3 3.5
Derivative Based Edge Detection
 From the previous slide, we can conclude that:

Local maxima of the absolute of the 1st derivative and Zero crossing
of the 2nd derivative occur at edges.
 Therefore, for detecting edges, we can apply zero crossing detection

to the 2nd derivative image or thresholding the absolute of the
1st derivative image.
 Nevertheless, derivative operator is very sensitive to noise as we

will see in the next slide.
Noisy Edges and Derivatives
Derivative operator is
f(x) df d2 f
a highpass filter and
dx dx 2 thus enhances noise.
AWGN s = 0.1
Edge responses
AWGN s = 1.0
are buried by
noise.
AWGN s = 10
Masks for Estimating Partial Derivatives
Normally, the mask for estimating partial derivative is anti-

symmetry with respect to the orthogonal axis
For example, the Sobel mask for
computing f is anti-symmetry
x
with respect to the y-axis. It has the
positive sign on the right side and
negative sign on the left side.
Notice that sum of all coefficients is

equal to zero to make sure that the
response of a constant intensity area
is zero.
Masks for Detecting Diagonal Edges
The mask for detecting -45-degree

edges is anti-symmetry with respect to
the –45-degree lines while the mask for
detecting 45-degree edges is
anti-symmetry with respect to the
45-degree lines.

Example of Image Gradient
f ( x, y ) f
y
f f f

x x y

Example of Image Gradient
f ( x, y ) f
y
f f f

x x y
Note: the original image is smoothed by a 5x5 moving average

mask first. (Images from Rafael C. Gonzalez and Richard E.
Example of Diagonal Edges
Using -45-degree mask Using 45-degree mask
Note: the original image is smoothed by a 5x5 moving average

mask first. (Images from Rafael C. Gonzalez and Richard E.
Laplacian Masks
The Laplacian masks are used to estimate the Laplacian image:
2P 2 P
 P 2  2
2
x y
Ideally, the Laplacian mask must be directional invariant: symmetry in

all direction (radially symmetry ). However, for 3x3 masks, there are
Only 8 possible directions. Hence, we can use the following masks:

Laplacian Masks
For a large scale Laplacian mask, we can use a Laplacian of Gaussian
(LOG) as a mask:
 x2  y2 
 x  y s 
2 2 2 
 2s
2


 G ( x, y )   
2
 e 
 s 4

Surface plot of
LOG, Looks like
a “Mexican hat” LOG image
Cross section 5x5 LOG

of LOG mask
Example of Laplacian Image
The angiogram image
(blood vessels)
P
We can compute the
Laplacian image by:
1. Smooth the image
by the Gaussian mask 1
2
2. Compute the Laplacian
image using the mask
Threshold of Zero crossing
The left image
2G * P

Pixel Oriented Image Segmentation: Thresholding
As a way to separate the elements of the picture by
the same properties of the pixel.
The only way to areas such as Intensity Thresholding.
1 f ( x, y )  T
g ( x, y )  
0 f ( x, y )  T
Works well with the subject and the background
clearly separate notice.
Histogram of the image is multimodal.
T = 102
500 color background
400 bacteria
300
200
100
0 50 100 150 200 250

After thresholding
Intensity Thresholding Example

Automatic Threshold Level Selection
The major problem of intensity thresholding is to find a good
threshold level
Algorithm: effective for bimodal histogram
1. Set initial value of T
2. T1  Average( p( x, y ) p( x, y )  T )
3. T2  Average( p( x, y ) p( x, y )  T )
T1  T2
4. T
2
5. Repeat step 2
Automatic Threshold Level Selection Example

Multilevel Intensity Thresholding
In some cases, an object with a different intensity
may require multiple Threshold Level.
Histogram
1500
T1 = 158
1000
T2 = 196
500
T3 = 228
0
0 50 100 150 200 250
T1< P <T2 T2< P <T3 P > T3

Noise Problem
When used with image noise. Pictures will be discrete.
peak unclear
Image degraded by Histogram
Gaussian noise (s =12) 250
T1 = 158
200
150
T2 = 196
100
50
T3 = 228
0
0 50 100 150 200 250
T1< P <T2 T2< P <T3 P > T3

Nonuniform Illumination Problem
An image can be expressed as
Reflectance
f ( x, y )  i ( x , y ) r ( x , y )
Function r(x,y)
i(x,y) = illumination component
Histogram r(x,y) = reflectance component
Illumination
Function i(x,y)
Image histogram
f(x,y)

Nonuniform Illumination and Global Thresholding
Global threshold level

Global thresholding of nonuniform
illumination image can cause huge
errors!
Histogram
Nonuniform illumination Global thresholding

image result
Nonuniform Illumination and Local Thresholding
Local thresholding:
1. Divide an image into subimages.
2. Threshold each subimage independently
2.1 Compute histogram of each subimage and select a suitable
threshold value for each subimage
2.2 threshold each subimage using a threshold value in 2.1
2.3 Combine all local thresholding results
Error
16 subimages Result of local thresholding

Histogram of Subimages and Local Thresholding
If areas of object and
background are not
balanced, a histogram
will be unimodal.
If areas of object and

background are nearly
equal, a histogram will
be bimodal

Example of Nonuniform Illumination Problem
When the image brightness is not uniform over the image.
The results will be tremendous.
900
800
700
600
500
400
300
200
100
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
T=0.4
Error edges
ISAN-DSP GROUP
Optimum Thresholding
Background
p1(z) = PDF of object pixels
Object p2(z) = PDF of background pixels
Error due to background pixels E1 (T )   p ( z )dz


2
classified as object pixels is :

Error due to object pixels
classified as background pixels is: E2 (T )   p1 ( z )dz
T
Total error = E(T )  P2 E1 (T )  P1E2 (T )
P1 = Probability of occurrence of object pixels

P2 = Probability of occurrence of background pixels


Pixel Oriented Image Segmentation for Color Images
RGB, CMY color models:

Thresholding based on distance in the color space.
HSI color model:

Thresholding based on H and S component mainly.

Color Segmentation in HSI Color Space
Color image Hue
1 2
3 4
Saturation Intensity
Color Segmentation in HSI Color Space (cont.)
Binary thresholding of S component Product of 2 and 5
with T = 10%
5 6
Red pixels
7 8
Histogram of 6 Segmentation of (Images
red color
from Rafaelpixels
C. Gonzalez and Richard E.
Color Segmentation in HSI Color Space (cont.)
Color image Segmented results of red pixels

Color Segmentation in RGB Vector Space

1. Each point with (R,G,B) coordinate in the vector space represents

one color.
2. Segmentation is based on distance thresholding in a vector space
1 if D (c( x, y ), cT )  T
g ( x, y )  
0 if D (c( x, y ), cT )  T
cT = color to be segmented.
D(u,v) = distance function c(x,y) = RGB vector at pixel (x,y).
Example: Segmentation in RGB Vector Space
Color image
Reference color cT to be segmented

cT  average color of pixel in the box
Results of segmentation in
RGB vector space with Threshold
value
T = 1.25 times the SD of R,G,B values
In the box

Color Image Segmentation Example

Pixel Oriented Image Segmentation for Multispectral Images
The principle of pixel oriented image segmentation can be applied
to the image.multispectrum like painting by division or partition
in Feature space. The separate elements of the color of the pixel itself.
Image Domain Partition
Munich, photos from space Boundary
1
Blue
0.5
0
1
0.2 0.8
0.4 0.6
0.6
0.4
From www.jpl.nasa.gov/radar/ 0.8
0.2
sircxsar/munch.html 1 Green
Red
Feature Space (RGB)
Self Organizing Map for Color Image Segmentation
We can use the Self Organizing Map (SOM) to calculate the right colors to use.
In the separate components of color pixels.
The process for primary colors

1. The number of primary colors to break.
2. Set the default color for the main color.
3. Each pixel values (R, G, B) to calculate the distance between the primary colors.
Every color on the color of the pixel.
4. The main color for the distance in the lowest third (winning the primary colors)
primary colors are colors that are nearby.
The color of the pixel Adjust the color of the paint on the way to the color of the pixe
5. Back in the 3-4 until the color of the primary does not change much.
The process of separating elements

1. Each pixel values (R, G, B) to calculate the distance between the primary colors.
Every color on the color of the pixel.
2. In the main color for the lowest one (primary colors to win) the pixels were
among the winners of the primary colors.
Self Organizing Map for Color Image Segmentation
Munich, contains images The results of the separate components by colo

White-space building in the city. Pictures of the SOM
Purple jungle area
Green farmland density
Black Sea area
Coloring sparse farmland
ISAN-DSP GROUP
Advantages and Disadvantage of Pixel Oriented Image Segmentation
Advantages
Because this method uses the properties of a single pixel.
Not much needs to be calculated. And spent a disadvantage
Do not apply to image noise. And the brightest
The results are often inconsistent and incoherent picture.
ISAN-DSP GROUP
Region Oriented Image Segmentation
As a way to separate the components of the image based on
the position of the pixels and the same.
Features of the pixel area If adjacent pixels and features are identical.
To the same group
The advantage of doing this is an area where continued.
Example Region oriented image segmentation.

1. Region Growing
2. Region Splitting and Merging
P
Connected
pixels
Pixel P and
its neighbors
Region Growing Algorithm
From seed pixel by pixel to the group.

Classes are homogenous areas grew.
Area expanded
Work will stop, can not expand out.

Possible over (On the edges of the object).
ISAN-DSP GROUP
Region Growing Image Segmentation Example
Region growing by example
MLB combination
If the intensity of adjacent pixels.
Light of the current maximum of four pixels.
Arranged in the same
Begin
from within
finished
the edge
ISAN-DSP GROUP
Histogram
X-ray image of defective weld


Region Splitting and Merging Algorithm
The process has two phases.
1. divided into sub-images, each sub-image.
The pixels within the uniform property
2. The overall picture is divided into groups
based on as well as the properties of the pixel
Region Splitting
1. Initial divided into sub-images.
2. For each pixel of the image if there

is no space inside the property. One
to divide indefinitely.
ISAN-DSP GROUP
Quadtree for Region Splitting Representation

Region Splitting Algorithm
Region Splitting
ISAN-DSP GROUP
Region Splitting Example
original size
The results of the Region Spliting.

standard deviation of no more than five sub-pixels
in the image.
ISAN-DSP GROUP
Region Merging Algorithm
2. Merging. Image sub-classified all the pixels within
the same features together.
ISAN-DSP GROUP
Region Merging Example
original size
The results of the Region Merging.

The average sub-pixel in the image is a combination
of not more than 50.
ISAN-DSP GROUP
Region Splitting and Merging Example

Obstacle of Region Oriented Image Segmentation
Problems of the Region Growing.
1. The criteria used to determine whether the pixel. "Homogenous"

In order to group together.
2. Determination seed pixel is very important. Must be defined
within the area decomposition
3. The rules for the program to stop working on the expansion.
To the edges of the object
Problems of the Region Splitting and Merging.

1. The criteria used to determine whether the total pixels in each image.
Featuring the same one or not.
2. To manage the complex data structures with sub-divided.
ISAN-DSP GROUP
Edge Oriented Image Segmentation
As a way to separate the discrete elements of the live image.
(Discontinuity) of features of the pixels at the interface between
the subject and the background. This approach is aimed primarily
at the edges of objects.
The end result of this approach is on the borderline between different areas
The continuation of
Color stitching
ISAN-DSP GROUP
Smoothed Step Edge and Its Derivatives
1.2
Intensity
1 Edge Edge
0.8
Gray level profile 0.6
0.4
0.2
0
-0.2
-1.5 -1 -0.5 0 0.5 1 1.5 2 2.5 3 3.5
0.06 Minimum
0.04
0.02
point
The 1st derivative 0
-0.02
-0.04 Maximum
-0.5point
-0.06
-1.5 -1 0 0.5 1 1.5 2 2.5 3 3.5
x 10 -3 Zero
5
4
3
crossing
2
1
+ +
The 2nd derivative 0
-1 - -
-2
-3
-4
-5
-1.5 -1 -0.5 0 0.5 1 1.5 2 2.5 3 3.5
Derivative Based Edge Detection
 From the previous slide, we can conclude that:

Local maxima of the absolute of the 1st derivative and Zero crossing
of the 2nd derivative occur at edges.
 Therefore, for detecting edges, we can apply zero crossing detection

to the 2nd derivative image or thresholding the absolute of the
1st derivative image.
 Nevertheless, derivative operator is very sensitive to noise as we

will see in the next slide.
Masks for Estimating Partial Derivatives
Normally, the mask for estimating partial derivative is anti-

symmetry with respect to the orthogonal axis
For example, the Sobel mask for
computing f is anti-symmetry
x
with respect to the y-axis. It has the
positive sign on the right side and
negative sign on the left side.
Notice that sum of all coefficients is

equal to zero to make sure that the
response of a constant intensity area
is zero.
Gradient Image
Operator associated with the first derivative of the two-dimensional
image is the Gradient Operator (Ñ). Which can be used to detect
edges. 2
P ˆ P ˆ  P   P 
2
P  i Gradient
j P      
x y Magnitude  x   y 
้ ้น P P P
ภาพตังต P
x y
Gradient Vector Field
Gradient Vector is a quantity that describes
the size and orientation of the object.
้ ้น
ภาพตังต
Gradient P ้ วของ
พืนผิ P
Vector P บริเวณขอบมีลกั ษ
Field เหมือนสันเขา ตาแหน
ขอบคือสันเขา
Gradient Based Image Segmentation
้ ้น
P
P|
Thresholding
ใช ้ T มีคา่ สูงไ
จะได ้ขอบทีข ่
Edge map บริเวณมุม
ใช ้ T มีคา่ ต่าไป
จะได ้ขอบทีหนา่ T=60 T=100
Laplacian Image
่ ยวข
Operator ทีเกี ่ ้องกับอนุ พน
ั ธ ์อันดับ 2 ของภาพ 2 มิต ิ คือ Gradient
Operator ( 2 )
่
ซึงสามารถใช ้ในการตรวจจั บขอบได ้
 2
P  2
P
2 P  2  2
x y
้ ้น P
ภาพตังต 2 P 2 P
2 P
x 2 y 2
Laplacian Image
่
เครืองหมายของ ่
เครืองหมายของ
 2 P เป็ นลบ 2 P เป็ นบวก
ภายนอกวัตถุ ภายในวัตถุ
้ ้น P
ภาพตังต 2 P
้
พืนผิวของP
 2
Laplacian Based Image Segmentation
้ ้น
2 P 2 P
Zero
Crossing
Detection Zero crossing
Detection ทาได ้
โดยทา thresholding
Edge map กับ 2 P โดยใช ้ T=0
Laplacian
ขยายสัญญาณรบกวน
ให ้ช ัดเจนขึน้ ISAN-DSP GROUP
Gradient VS Laplacian
Gradient Laplacian
P 2P
ผลตอบสนองต่อขอบ Local Extrema Zero Crossing
แบบขัน้ of |P|
(step edge) ของวัตถุ
ความสาคัญ บอกถึงทิศทาง บอกถึงตาแหน่ ง
และขนาดของ ของขอบของวัตถุ
ขอบของวัตถุ
การตรวจจับขอบ Thresholding Zero Crossing
|P| Detection
จุดอ่อน ผลลัพธ ์มักได ้ มักไม่ทนทานต่อ
่
edge map ทีไม่ สัญญาณรบกวน
สมบูรณ์
ISAN-DSP GROUP
Boundary Extraction Based on Particle motion in a Vector Field
Gradient field Tangential velocity field
Gradient Vector Field Edge Vector Field

(Hamiltonian Gradient Field)
P ˆ P ˆ P ˆ P ˆ
P  i j P   i  j
x y y x
Edge Vector Field
Tangential velocity field
Edge vector field

มีลก
ั ษณะเหมือนกระแสน
P ˆ P ˆ ไหลวนรอบวัตถุ
P   i  j
y x
ISAN-DSP GROUP
Particle Trajectory in an Edge Vector Field
Tangential velocity field
ตาแหน่ งของอนุ ภาคคานวณได ้จาก
pk  pk 1   ek 1
pk ตาแหน่ งปัจจุบน ่
ั ของอนุ ภาคทีเว
่
pk 1 ตาแหน่ งในอดีตของอนุ ภาคทีเว
ek 1 edge vector field ณ ตาแหน
ที่ pk 1
 ค่าคงที่  > 0
่ ้นของอนุ ภาค
ตาแหน่ งเริมต
ผลลัพธ ์: ใน edge vector field อนุ ภาคโคจรรอบวัตถุมรี ะยะห่างจา
้ อยๆ
ขึนเรื ่ จนเกิดทางเดินรูปก ้นหอย (spiral trajectory)
Example: Particle Trajectory in an Edge Vector Field
้ ้น
ภาพตังต P
P0
Edge vector field ทางเดินของอนุ ภาค

dient field Sign of Laplacian image Normal compressive velocity field
A Normal Compressive Vector Field
่ จะท
เพือที ่ าใหอ้ นุ ภาคเคลือนที ่ ่
ไปตามขอบของวั ตถุโดยไม่หลุดไป
วัตถุ เราต ้องการ vector field อีกสนามทีมี ่ คณุ สมบัต ิ “บีบอัด” (nor
compressive) เพือบั ่ งคับใหอ้ นุ ภาคเดินตามขอบ
ตัวอย่าง Normal Compressive Field
1
n  P 2 P
c
P 2 P n
Example: Normal
A Normal Compressive
compressive velocity field
Vector Field
1
n  P 2 P
c
ISAN-DSP GROUP
Combined Vector Field
t+n
v  e   n
v มีลก
ั ษณะลูเ่ ข ้าสูข
่ อ
และไหลวนรอบวัตถุ
ISAN-DSP GROUP
Particle Trajectory in a Combined Vector Field
ตาแหน่ งของอนุ ภtาค

p+ nk  pk 1   ek 1   nk 1
อนุ ภาคโคจรรอบวัตถุตาม
เส ้นทางเดียว
เราสามารถแกะรอยเ
ขอบของวัตถุได ้จากทาง
ของอนุ ภาค
่ ้นของอนุ ภา
ตาแหน่ งเริมต
Example: Particle Trajectory in a Combined Vector Field
Edge vector field 1

n  P 2 P
c
Particle trajectory
P0 pk  pk 1   ek 1   nk 1
Shape distortion at a corner

Boundary Extraction of Multiple Objects
1-way boundary extraction 2-way boundary extraction
ถ ้าใหอ้ นุ ภาคเดินไปในทิศทางเดียว
่ สมบูรณ์
จะมีโอกาสได ้เส ้นขอบทีไม่ ISAN-DSP GROUP
Boundary Extraction Algorithm Based on Particle
Motion in a Vector Field
่
1. เลือก Mask ทีเหมาะสมในการค านวณ Gradient และ Laplacian
ของภาพ
2. คานวณ edge vector field และ normal compressive vector fie
่ ้นใหอ้ นุ ภาค ณ ตาแหน่ ง Local maxima
3. กาหนดตาแหน่ งเริมต
ของ
Gradient magnitude
4. แกะรอยเส image นของอนุ ภาค
้นขอบตามทางเดิ
่
5. หยุดการทางานเมืออนุ ภาคเดินกลับมาทีจุ่ ดเริมต
่ ้นหรือพบกับเส ้น
่
ขอบทีเคยผ่ านมาแล ้ว
ISAN-DSP GROUP
Boundary Extraction Example
(Obtained using multiple scale Gaussian differential

masks with smin = 0.5, smax = 2.0)
ภาพจาก The whole brain atlas, www.med.harvard.edu/AANLIB/home.html ISAN-DSP GROUP

Boundary Extraction Example: Flaw Detection
Original X-ray image Gradient image
Boundary extraction results (s=1.0)

Flaws
Multiscale Boundary Extraction
Gradient images and extracted boundaries
s1=0.5 s2=4.77 s3=18.81

Conclusions
่ ประสิทธิภาพ จะต ้องมีคณ

Image segmentation ทีมี ุ สมบัตส
ิ า
1. มีขบวนการปรับปรุงภาพทีดี่ เพือท
่ าให ้ภาพง่ายต่อการแยกองค ์ป
2. ใช ้ข ้อมูลสาคัญให ้เป็ นประโยชน์
2.1 คุณสมบัตท ี่ อนกันของพิกเซลภายในพืนที
ิ เหมื ้ เดี่ ยวกัน
2.2 คุณสมบัตท ี่
ิ แตกต่ างกันของพิกเซลในพืนที ้ ที่ ต่
่ างกัน
2.3 ตาแหน่ งของพิกเซล และการเชือมต่ ่ อกัน
2.4 ตาแหน่ ง ขนาด ทิศทางของรอยต่อระหว่างพืนที ้ ่
2.5 สมมุตฐิ านเบืองต ้ ้นทีจ่ าเป็ น (ความรู ้ชันสู
้ ง)
่
3. มีโครงสร ้างข ้อมูลทีสามารถรองรั บวิธก
ี ารแยกองค ์ประกอบภาพไ
่ ดเท่าทีจ่ าเป็ น
4. ต ้องการ user defined parameters น้อยทีสุ
ISAN-DSP GROUP
Edge Linking

Chapter10 Segmentation PDF

Încărcat de

Informații document

Titlu original

Drepturi de autor

Formate disponibile

Partajați acest document

Partajați sau inserați document

Opțiuni de partajare

Vi se pare util acest document?

Este necorespunzător acest conținut?

Drepturi de autor:

Formate disponibile

Chapter10 Segmentation PDF

Încărcat de

Drepturi de autor:

Formate disponibile

Image Segmentation

Element of Image Analysis

High level process

 One application of spatial filtering is shape detection: finding

 Unlike frequency selective masks that are designed based

 A mask for shape detection usually contains the shape or a part

 The location that is most correlated to the mask is the

 Notice that sum of all coefficients of the mask is equal to zero.

 Similar to point detection, line detection can be performed

Line detection masks

Notice that –45 degree

(Images from Rafael C.

Generally, objects and background have different intensities. Therefore,

(Images from Rafael C. Gonzalez and Richard E.

(Images from Rafael C. Gonzalez and Richard E.

 From the previous slide, we can conclude that:

 Therefore, for detecting edges, we can apply zero crossing detection

 Nevertheless, derivative operator is very sensitive to noise as we

Normally, the mask for estimating partial derivative is anti-

Notice that sum of all coefficients is

The mask for detecting -45-degree

(Images from Rafael C. Gonzalez and Richard E.

(Images from Rafael C. Gonzalez and Richard E.

Note: the original image is smoothed by a 5x5 moving average

Using -45-degree mask Using 45-degree mask

Note: the original image is smoothed by a 5x5 moving average

Ideally, the Laplacian mask must be directional invariant: symmetry in

(Images from Rafael C. Gonzalez and Richard E.

Cross section 5x5 LOG

(Images from Rafael C. Gonzalez and Richard E.

0 50 100 150 200 250

(Images from Rafael C. Gonzalez and Richard E.

(Images from Rafael C. Gonzalez and Richard E.

0 50 100 150 200 250

T1< P <T2 T2< P <T3 P > T3

0 50 100 150 200 250

T1< P <T2 T2< P <T3 P > T3

(Images from Rafael C. Gonzalez and Richard E.

Global threshold level

Nonuniform illumination Global thresholding

16 subimages Result of local thresholding

If areas of object and

(Images from Rafael C. Gonzalez and Richard E.

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

Error due to background pixels E1 (T )   p ( z )dz

Total error = E(T )  P2 E1 (T )  P1E2 (T )

P1 = Probability of occurrence of object pixels

(Images from Rafael C. Gonzalez and Richard E.

(Images from Rafael C. Gonzalez and Richard E.

RGB, CMY color models:

HSI color model:

(Images from Rafael C. Gonzalez and Richard E.

Color image Segmented results of red pixels

(Images from Rafael C. Gonzalez and Richard E.

(Images from Rafael C. Gonzalez and Richard E.

1. Each point with (R,G,B) coordinate in the vector space represents

Reference color cT to be segmented

(Images from Rafael C. Gonzalez and Richard E.

(Images from Rafael C. Gonzalez and Richard E.

The process for primary colors

The process of separating elements

Munich, contains images The results of the separate components by colo

Example Region oriented image segmentation.