Loughborough University
Browse

Anisotropic mean shift based fuzzy c-means segmentation of deroscopy images

Download (666.19 kB)
journal contribution
posted on 2014-06-04, 09:17 authored by Huiyu Zhou, Gerald SchaeferGerald Schaefer, A.H. Sadka, M.E. Celebi
Image segmentation is an important task in analysing dermoscopy images as the extraction of the borders of skin lesions provides important cues for accurate diagnosis. One family of segmentation algorithms is based on the idea of clustering pixels with similar characteristics. Fuzzy c-means has been shown to work well for clustering based segmentation, however due to its iterative nature this approach has excessive computational requirements. In this paper, we introduce a new mean shift based fuzzy c-means algorithm that requires less computational time than previous techniques while providing good segmentation results. The proposed segmentation method incorporates a mean field term within the standard fuzzy c-means objective function. Since mean shift can quickly and reliably find cluster centers, the entire strategy is capable of effectively detecting regions within an image. Experimental results on a large dataset of diverse dermoscopy images demonstrate that the presented method accurately and efficiently detects the borders of skin lesions.

History

School

  • Science

Department

  • Computer Science

Citation

ZHOU, H. ... et al., 2009. Anisotropic mean shift based fuzzy c-means segmentation of deroscopy images. IEEE Journal of Selected Topics in Signal Processing, 3 (1), pp. 26 - 34.

Publisher

© IEEE

Version

  • AM (Accepted Manuscript)

Publication date

2009

Notes

This article was published in the IEEE Journal of Selected Topics in Signal Processing [© IEEE] and the definitive version is available at: http://dx.doi.org/10.1109/JSTSP.2008.2010631 Personal use of this material is permitted. Permission from IEEE must be obtained for all other users, including reprinting/ republishing this material for advertising or promotional purposes, creating new collective works for resale or redistribution to servers or lists, or reuse of any copyrighted components of this work in other works.

ISSN

1932-4553

Language

  • en

Usage metrics

    Loughborough Publications

    Exports

    RefWorks
    BibTeX
    Ref. manager
    Endnote
    DataCite
    NLM
    DC