Local and Global Measure of Dissimilarity between Two Segmentations
A. Merdani1, A. Kharbach2, M. Rahmoun3, B. Bellach4, M. Elayachi5, M. Elhitmy6
1Amar Merdani, Department of Electronic Systems, Computer Image, Mohammed I, National University of Applied Sciences, Laboratory, Oujda Morocco.
2Amina Kharbach, Department of Electronic Systems, Computer Image, Mohammed I, National University of Applied Sciences, Laboratory, Oujda Morocco.
3Mohammed Rahmoun, Department of Electronic Systems, Computer Image, Mohammed I, National University of Applied Sciences, Laboratory, Oujda Morocco.
4Benaissa Bellach, Department of Electronic Systems, Computer Image, Mohammed I, National University of Applied Sciences, Laboratory, Oujda Morocco.
5M. Elayachi, Department of Electronic Systems, Computer Image, Mohammed I, National University of Applied Sciences, Laboratory, Oujda Morocco.
6Mohammed Elhitmy, Department of Electronic Systems, Computer Image, Mohammed I, National University of Applied Sciences, Laboratory, Oujda Morocco.
Manuscript received on 12 November 2014 | Revised Manuscript received on 22 November 2014 | Manuscript Published on 30 November 2014 | PP: 11-14 | Volume-4 Issue-6, November 2014 | Retrieval Number: F1846114614/14©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: The implementation of a segmentation method in a system requires knowledge of the performance of the method in a given situation. Hence, it is highly desirable to have a criterion for measuring the quality of the result obtained by a segmentation algorithm. This study focuses on two measures of dissimilarity between two segmentations, by means of a mapping. The local measure proposed is based on the map of local dissimilarities that capture the differences between two images. This allows a simple way to quantify the local dissimilarities and to determine their spatial distribution. Thus, we are building a global measure based on local measurements. Both measures local and global are successfully tested on synthetic and medical images.
Keywords: K-Means, Region Growing, Hausdorff Distance, Distance Transformation, Local Dissimilarity, Global Dissimilarity.
Scope of the Article: Measurement & Performance Analysis