A Comparative Analysis of Segmentation Techniques to Extract Skin Lesion Regions
A. Ranichitra1, D. Seethalakshmi2
1Dr A. Ranichitra, Department of Computer Science, Sri S.Ramasamy Naidu Memorial College, Sattur, India.
2D. Seethlakshmi, Department of Computer Science, Sri S.Ramasamy Naidu Memorial College, Sattur, India.
Manuscript received on 10 August 2017 | Revised Manuscript received on 18 August 2017 | Manuscript Published on 30 August 2017 | PP: 95-100 | Volume-6 Issue-6, August 2017 | Retrieval Number: F5128086617/17©BEIESP
Open Access | Editorial and Publishing Policies | Cite | Mendeley | Indexing and Abstracting
© The Authors. Blue Eyes Intelligence Engineering and Sciences Publication (BEIESP). This is an open access article under the CC-BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/)
Abstract: Skin diseases are the most common health problems in worldwide. Diagnosis of the skin disease depends on the extraction of the abnormal skin region. In this paper, Segmentation techniques to extract the skin lesion regions are proposed and their results are compared based on the statistical and texture properties. The acquired skin images are preprocessed by median filter and segmented by Edge-based segmentation, Morphological segmentation and K-means clustering techniques. The statistical features mean and standard deviation and the texture features contrast and energy are calculated for all the segmented skin lesion images. The performance of the three segmentation techniques are compared and found that the K-Means algorithm yields better results without any over and under segmentation.
Keywords: Edge Based Segmentation, Energy, Contrast, KMeans Clustering, Morphological Segmentation, Mean, Standard Deviation, Skin Lesion,
Scope of the Article: Clustering