Scribd
Upload
24 views

DigitalImageProcessing BorderDetectionAndLabeling

The document describes an algorithm to detect borders and label connected pixels in an image using 4-neighborhood connectivity. It starts with an example image and runs the algorithm in two phases, first labeling pixels and then correcting mislabeled pixels, to uniquely label each connected component.

Uploaded by

Thuy Linh
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
24 views

DigitalImageProcessing BorderDetectionAndLabeling

The document describes an algorithm to detect borders and label connected pixels in an image using 4-neighborhood connectivity. It starts with an example image and runs the algorithm in two phases, first labeling pixels and then correcting mislabeled pixels, to uniquely label each connected component.

Uploaded by

Thuy Linh
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 5

stanbul Teknik niversitesi

Elektrik Elektronik Fakltesi


Telekomnikasyon Mhendislii
Blm

TEL 432 E

DIGITAL IMAGE

PROCESSING

-Homework 1-

Batuhan Osmanoglu
040010250
Border Detection And Labeling using 4-Neighborhood

It is asked to develop an algorithm that gives a unique label to different groups of


cells. Groups are defined using 4-Neighborhood concept.

4-Neighborhood:

In this concept every cell can be connected with its 4 neighbors. These neighbors
might be located on the left, right hand sides and above or below the pixel itself.

P(x-1, y-1) P(x, y-1) P(x+1, y+1) List of 4-Neighborhood pixels of


Pixel (x, y):
{P(x, y-1), P(x, y+1), P(x-1, y),
P(x-1, y) P(x, y) P(x+1, y) P(x+1, y)}

P(x-1, y+1) P(x, y+1) P(x+1, y+1)

To employ this concept and give unique labels to different group of neighbor pixels a
MatLab code might be used as given below:

%Simple Detect
clc;
colormap(bone);
%Create Image Array 8 X 13
Im=[ 0,0,0,0,0,0,0,0,0,0,0,0,0;
0,0,1,0,0,0,0,0,0,1,0,1,0;
0,0,0,0,0,0,0,0,0,0,0,0,0;
0,1,1,1,0,1,1,1,0,1,0,1,0;
0,0,1,0,0,0,1,0,0,1,0,1,0;
0,0,1,0,0,0,1,0,0,1,0,1,0;
0,1,1,1,0,0,1,0,0,1,1,1,0;
0,0,0,0,0,0,0,0,0,0,0,0,0;
];
%Check Connectivitiy, If an unconnected pixel is found, create a new label!

%lc=2; %label color, Actually no need to set this up. Just to be on the safe side.
lcmax=2;
[nr,nc]=size(Im); %number of rows and columns
im_out=Im;
for r=1:nr-1
for c=1:nc-1
if im_out(r,c)==1
lcmax=lcmax+1;
%set a lcmax value to achieve label uniqueness...lcmax=maximum label value ever used
lc=lcmax;
im_out(r,c)=lc;
elseif im_out(r,c) > 1
lc = im_out(r,c);
elseif im_out(r,c)==0
continue;
else
break;
end
if im_out(r,c+1)==1 %right neighbor
im_out(r,c+1)=lc;
end
if im_out(r+1,c)==1 %bottom neighbor
im_out(r+1,c)=lc;
end
end
end
% Do some correction. In other words, PHASE 2 :)
for r=nr:-1:2 %This time go in the reverse direction.
for c=nc:-1:2
if im_out(r,c)== 0
continue;
end
if (im_out(r,c)~= im_out(r,c-1)) *(im_out(r,c-1)>0) % incase there is a mislabeled
neighbor re-label it. (Left neighbor)
im_out(r,c-1) = im_out(r,c);
end
if (im_out(r,c)~= im_out(r-1,c))* (im_out(r-1,c)>0) % incase there is a mislabeled
neighbor relabel it. (Upper neighbor)
im_out(r-1,c) = im_out(r,c);
end
end
end
%Output results
im_out
figure(1);
imagesc(im_out);

Algorithm:

First an image is formed and given the name Im.


Figure 1: Screen capture of command: imagesc(Im)

Once it is created algorithm needs to scan it for two times, to obtain unique group
labels. The first pass labels each cell trying to maintain uniqueness, however, because of
its one-way operation it does not go back and correct mislabeled pixels. Therefore, a
second pass over the output image is needed to correct mislabeled pixels and decrease the
number of groups to optimum level.

Below one can find values for original image, output of first phase and second
phase.
Original Image:

Im =
0 0 0 0 0 0 0 0 0 0 0 0 0
0 0 1 0 0 0 0 0 0 1 0 1 0
0 0 0 0 0 0 0 0 0 0 0 0 0
0 1 1 1 0 1 1 1 0 1 0 1 0
0 0 1 0 0 0 1 0 0 1 0 1 0
0 0 1 0 0 0 1 0 0 1 0 1 0
0 1 1 1 0 0 1 0 0 1 1 1 0
0 0 0 0 0 0 0 0 0 0 0 0 0
Output of Phase 1:
0 0 0 0 0 0 0 0 0 0 0 0 0
0 0 3 0 0 0 0 0 0 4 0 5 0
0 0 0 0 0 0 0 0 0 0 0 0 0
0 6 6 6 0 7 7 7 0 8 0 9 0
0 0 6 0 0 0 7 0 0 8 0 9 0
0 0 6 0 0 0 7 0 0 8 0 9 0
0 10 6 6 0 0 7 0 0 8 8 9 0
0 0 0 0 0 0 0 0 0 0 0 0 0

Output of Phase 2:
im_out =
0 0 0 0 0 0 0 0 0 0 0 0 0
0 0 3 0 0 0 0 0 0 4 0 5 0
0 0 0 0 0 0 0 0 0 0 0 0 0
0 6 6 6 0 7 7 7 0 9 0 9 0
0 0 6 0 0 0 7 0 0 9 0 9 0
0 0 6 0 0 0 7 0 0 9 0 9 0
0 6 6 6 0 0 7 0 0 9 9 9 0
0 0 0 0 0 0 0 0 0 0 0 0 0

It is easily seen that there are no pixels labeled 8 or 10 in the last output.

Figure 2: Output of command: colormap(bone); imagesc(im_out);

You might also like