Paper Info
Title
Minimum Cross Entropy Thresholding Using Entropy-Li Based on Log-normal Distribution for Skin Cancer Images
Abstract
Accurate thresholding of skin cancer images is avery essential issue in medical diagnostic applications andbiometric authentication and verification systems [1], anotherimportant issue in such applications relates to itscomputational complexity. The objective of this study is todevelop a more accurate and faster solution for estimating theoptimal thresholding value of skin images. This work proposesa new algorithm for an Iterative Minimum Cross EntropyThresholding method based on Log-normal distribution(MCET-Lognormal), under the assumption that the data ofskin images is best modelled as a mixture of Log-normaldistributions. The proposed method was applied on bi-modalskin cancer images and promising experimental results wereobtained. Evaluation of the resulting segmented skin imagesshows that the proposed method yields better estimation of theoptimal threshold than does the same MCET method withGaussian distribution (MCET-Gaussian). It also reducescomputational time compared to sequential search techniques.
Year
DOI
Venue
2011
10.1109/SITIS.2011.86
SITIS
Keywords
Field
DocType
log-normal distribution,skin cancer images,anotherimportant issue,skin image,proposed method yield,mcet method withgaussian distribution,accurate thresholding,skin cancer image,segmented skin imagesshows,minimum cross entropy thresholding,avery essential issue,cross entropy,sequential search,iterative algorithm,cancer,log normal distribution,image segmentation,entropy,iterative methods,computational complexity,skin
Computer vision,Pattern recognition,Medical diagnostic,Computer science,Iterative method,Image segmentation,Artificial intelligence,Minimum cross entropy,Thresholding,Linear search,Log-normal distribution,Computational complexity theory
Conference
Citations 
PageRank 
References 
0
0.34
5
Authors
3
Name
Order
Citations
PageRank
Duaa H. AlSaeed121.52
Ali El Zaart2528.17
Ahmed Bouridane383799.53