Image Processing Reference
In-Depth Information
11
Rough Fuzzy Measures in Image
Segmentation and Analysis
11.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
11-1
Motivation Chapter Organization
11.2 Clustering Based Image Segmentation . . . . . . . . . . . . .
11-3
Overview of Segmentation Methods Image Clustering
11.3 Image Segmentation Evaluation . . . . . . . . . . . . . . . . . . . . .
11-4
Segmentation Quality Measures Cluster Validity
Measures
11.4 RECA Algorithms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
11-6
Rough Set Theory Generalized Rough Set Theory
Crisp-Crisp Distance RECA Fuzzy-Crisp Difference
RECA Fuzzy-Fuzzy Threshold RECA Fuzzy-Fuzzy
Difference RECA
11.5 Experimental Setup . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11-12
Images Populations Standard Indices - SI
Parameters for RECA Measures
11.6 Experimental results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11-15
The correlation between standard and rough measures
The Best Solutions Population Indices
Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11-21
Acknowledgments
Dariusz Malyszko
Bialystok University of Technology
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11-24
Jaroslaw Stepaniuk
Bialystok University of Technology
Bibliography
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11-25
11.1
Introduction
Image segmentation presents extremely important routine with many applications in image
processing (Gonzales and Woods, 2002; Haralick and Shapiro, 1985). The issue of correct
image segmentation definition, creation and evaluation still remains unsolved and most of-
ten the solution to these problems mainly depends on the predefined goal. High quality and
robustness of image segmentation routines is of considerable importance in the practical ap-
plications and at the same time contemporary research. Image segmentation makes possible
further higher level image processing such as feature extraction, pattern recognition, and
classification. Despite the fact that much effort has been put into elaboration of suitable
segmentation algorithms the problem is still open in many areas that require particular seg-
mentation characteristics and continuous development of new imagery technologies makes
the segmentation quality issue constantly more pressing.
Image segmentation presents active research subject in the last decades. High quality
image segmentation routines are of great importance in practical applications owing their
11-1
 
 
 
 
 
 
 
 
 
 
Search WWH ::




Custom Search