IMAGE SEGMENTATION USING

DBSCAN

Supervised By:- Presented By:-

Dr. S. S. J. Nanda Rohitash Verma
Assistant Professor M.Tech ECE
MNIT Jaipur 2018PEC5174
Objective:-

Take Minimum 20 similar kind of Image and apply DBSCAN

Algorithm to them.
 Calculate DBSCAN Parameter.
Calculate Statistical Parameter Of all Images.
Apply these Statistical Parameter to train neural network.
After Training test neural network.
 Compare DBSCAN parameter with obtained DBSCAN parameter
from Neural network.
Contents:-
Image segmentation
Image Segmentation Method
DBSCAN Algorithm
Algorithm Steps for DBSCAN Clustering
Results
Advantage of DBSCAN
Disadvantage of DBSCAN
Neural Network Output
Conclusion
References
Image Segmentation:-

Process of subdividing an image.

Main purpose of subdividing an image into its constituent parts or
objects present in the image is that we can further analyze each of
these constituents or each of the objects present in the image.
These constituents can be analyzed to extract some information so that
those information are useful for high level machine vision
applications.
Naturally level of Segmentation is application depended.
Image Segmentation Method:-

a) Discontinuity based approach:-

 In this method subdivision of image is carried out based on some
abrupt changes in gray levels of image.
 Isolated points
 Lines present in the image
Identification of edge
b) Similarity based approach:-

We try to group those pixel in image which are similar in some
sense.
Thresholding operation.
Region growing based approach.
Region splitting and merging.
DBSCAN Algorithm:-

Relies on a density based notion of clusters.

Clusters are identified by looking at the densities of points.
Regions with a high density of points depict the existence of clusters
whereas regions with a low density of points indicate clusters of noise
or cluster of outliers.
This algorithm is particularly suited to deal with large dataset with
noise.
It is useful to identify clusters with different size and shape.
It uses the concept of density reachability and density connectivity
Density Reachability:-

A point “p” is said to be density reachable from a point “q” if point

“p” is within distance from point “q” and “q” has sufficient number of
points in its neighbor which are within distance .
Density Connectivity:-

A point “p” and “q” are said to be density connected if there exist a
point “r” which has sufficient number of points in its neighbours and
both the point “p” and “q” are within the distance.
DBSCAN Flow graph:-
Algorithm Steps for DBSCAN Clustering:-

Let X={X1,X2,X3------XN} be the set of data points.

DBSCAN requires two parameters :(eps) and minimum number of
points require to form a cluster(minpts).
1). Start with an arbitrary starting point.
2). Extract the neighborhood of this point using .
3). If there are sufficient neighborhood around this point then clustering
process starts and point is marked as visited else this point is labeled as
noise(Later this point can become the part of the cluster).
4). If a point is found to be a part of the cluster then its neighborhood is
also the part of the cluster and the above procedure from step 2 is
repeated for all neighborhood points. This is repeated until all points in
the cluster is determined.
5). A new unvisited point is retrieved and processed, Leading to the
discovery of a further cluster or noise.
6). This Process continues until all points are marked as visited.
DBSCAN Algorithm
Advantage Of DBSCAN

 Does not require a priori specification of number of clusters.

Able to identify noise data while clustering.
 DBSCAN is great to separating high density clusters versus low
density clusters in given dataset.
Disadvantage of DBSCAN

DBSCAN fails in case of varying density clusters.

Sensitive to clustering parameter eps and minpts.
Images after applying DBSCAN:-
DBSCAN Parameter:-
Image NAme Min.pts Radius Accuracy
Sudan flood 4 0.66 97.27%
South East india 3 0.8 99.21%
Flood

Colombia 3 0.46 `100%

Uk flood 3 0.765 100%
South Asia Flood 3 0.796 99.60%
Kerla Flood 4 0.4 92.02%
Morganza Flood 4 0.5 96.48%
Souris River 5 0.61 93.36%
Flood

Missisipi River 4 0.8 98.44%

Flood

Pakistan Flood 4 0.7 91.80%

Dakota Flood 4 0.55 95.31%
Burma Flood 4 0.7 93.75%
Bihar Flood 4 0.5 99.61%
Carolina Flood 3 0.46 100%
St.John River 4 0.62 97.24%
Flood
Neural Network Output:-

Red Curve-Desired Output. 1

Blue Curve-Tested Output. 0.8
0.6

0.4

0.2

-0.2

-0.4

-0.6
0 1 2 3 4 5 6 7 8 9 10
Error Graph:-

10 -3
16

14

12

10

-2

-4
1 2 3 4 5 6 7 8 9 10
Conclusion:-

In this, a image segmentation by using DBSCN is presented.

Different value of eps and minpts are required for optimum image
segmentation.
Mean, Variance ,Entrophy and Crosscorrelation Coefficient Calculated
and applied to neural network for training.
After Training Neural Network is tested on Similar Kind Of image.
References:-
1). Qixiang Ye, Wen Gao, Wei Zeng, “Color image segmentation using density-
based clustering “, International Conference of Acoustics,Speech,and Signal
Processing , August 2003.
2).https://medium.com/@elutins/dbscan-what-is-it-when-to-use-it-how-to-use-it-
8bd506293818.
THANK U------