Abbas Cheddad, Joan Condell, Kevin Curran and Paul Mc Kevitt Intelligent Systems Research Centre Faculty of Computing and Engineering University of Ulster.

Abbas Cheddad, Joan Condell, Kevin Curran and Paul Mc Kevitt

Intelligent Systems Research Centre Faculty of Computing and Engineering

University of UlsterUnited Kingdom

Emails: cheddad-a [ AT ] email.ulster.ac.ukWeb: http://www.infm.ulst.ac.uk/~abbasc/

Presentation Outline

• Introduction

• Applications

• Our Method

• Examples

• Conclusions

IntroductionSegmentation techniques can be classified into two

categories: boundary-based techniques and region-based techniques.

Region-based algorithms include region growing, region splitting and region merging

k-means minimize the mean squared distance from each data point to its nearest center (k)

Dynamic thresholding determined by examining repetitively the minima between two peaks in the bi-model image histogram

Edge detection Sobel, Prewitt, Laplacian, and Canny

Voronoi Diagram (VD) based on selected feature points residing along the image edges of high gradient magnitude (M. A. Suhail et al. and M. Burge and W. Burger)

3

4

IntroductionApplications

Remote sensingVehicle and robot navigationMedical imagingOptical Character Recognition (OCR)SkeletonizationScene analysisShape reconstruction, etc

5

MethodologyImage segmentation remains a long-standing problem in

computervision and has been found difficult and challenging for two

mainreasons (Z. Tu and S. Zhu):

The fundamental complexity of modelling a vast amount of visual data that appears in the image is a considerable challenge

The intrinsic ambiguity in image perception, especially when it concerns the so-called unsupervised segmentation (e.g., a decision whereby a region cut is not a trivial task)

6

MethodologyVoronoi Diagram (VD): Given a set of 2D points, the Voronoi region for a point Pi is defined as the set of all the points that are closer to Pi than to any other points. The dual tessellation of VD is known as the Delaunay Triangulation (DT).

VD of four generators

VD of two generators

VD of three generators

7

Methodology

VD of n scattered generators

8

Methodology

Various literature studies have tended to apply VD on the image

itself (after binarizing it and capturing its edges). This is usually

time consuming

Thus, VD is constructed from feature generators that result from

gray intensity frequencies. O (n log n), where n<=255

10/04/23 9

Methodology

Local flip effect on the histogram

Voronoi Diagram in red and Delaunay Triangulations in blue applied on an image histogram

10

Examples

Examples

12

Conclusions

We have presented our novel algorithm for image segmentation based on points geometry derived from the image histogram

Our proposal shows less complexity while maintaining high performance

This work is a pre-processing phase for our ongoing research on adaptive digital image Steganography. The latter is the science of concealing confidential data in multimedia medium in an imperceptible way

Contacts

WWW: http://www.abbascheddad.net

Email: cheddad-a [ AT ] email.ulster.ac.uk