Open AccessGeneralised fuzzy c‐means clustering algorithm with local information
Author(s) -
Memon Kashif Hussain,
Lee DongHo
Publication year - 2017
Publication title -
iet image processing
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.401
H-Index - 45
eISSN - 1751-9667
pISSN - 1751-9659
DOI - 10.1049/iet-ipr.2016.0282
Subject(s) - cluster analysis , robustness (evolution) , fuzzy clustering , computer science , fuzzy logic , pattern recognition (psychology) , image segmentation , artificial intelligence , algorithm , data mining , segmentation , flame clustering , spatial analysis , cure data clustering algorithm , canopy clustering algorithm , noise (video) , mathematics , image (mathematics) , statistics , biochemistry , chemistry , gene
Much research has been conducted on fuzzy c‐means (FCM) clustering algorithms for image segmentation that incorporate the local neighbourhood information into their objective function in order to mitigate problems related to noise sensitivity and poor performance. Although the bias‐corrected FCM, FCM with spatial constraints, and adaptive weighted averaging algorithms have proven to be robust to noise for image segmentation using local spatial image information, they have some disadvantages: (i) they are limited to single feature input data (i.e. intensity level feature), (ii) their robustness to noise and effectiveness heavily depend on a crucial parameter α , and (iii) it is difficult to find the optimal value of α , which is generally selected experimentally. In this study, to overcome all of these disadvantages, the authors present a generalisation of these types of algorithms that is applicable to cluster M‐ features input data. The proposed generalised FCM clustering algorithm with local information (GFCMLI) not only mitigates the disadvantages of standard FCM, but also highly improves the overall clustering performance. Experiments have been performed on several noisy data and natural/real‐world images in order to demonstrate the effectiveness, efficiency, and robustness to noise of the GFCMLI algorithm as compared with conventional methods.