skin hair removal in dermoscopic images using soft color morphology - - PowerPoint PPT Presentation

skin hair removal in dermoscopic images using soft color morphology

• P. Bibiloni
• M. González-Hidalgo
• S. Massanet

AIME 2017, 23/JUN/2017

Department of Mathematics and Computer Science University of the Balearic Islands (UIB) Palma, Spain Soft Computing, Image Processing and Aggregation (SCOPIA) research group

motivation

1

• ur method

(a) Original image (b) Hairless image

How? Hair Detection + Inpainting

2

• ur method

(a) Original image (b) Hairless image

How? Hair Detection + Inpainting

2

hair detection method

Color Mathematical Morphology Extract thin and dark objects.

(a) Original (b) Contrast

enhanced image

(c) Hair Mask

3

hair detection method

Color Mathematical Morphology Extract thin and dark objects.

(a) Original (b) Contrast

enhanced image

(c) Hair Mask

3

inpainting method

Color Mathematical Morphology Recover information from surrounding objects.

(a) Hair mask (b) Hairless image

(with missing information!)

(c) Hairless image

4

inpainting method

Color Mathematical Morphology Recover information from surrounding objects.

(a) Hair mask (b) Hairless image

(with missing information!)

(c) Hairless image

4

conclusions

∙ Mathematical morphology for color images. ∙ Competitive results:

∙ hair removal vs. texture preservation.

∙ Preliminary step.

5

what next?

Figure: Annotation tool - Overview

6

what next?

Figure: Annotation tool - Detail

6

acknowledgments

This work was partially supported by the project TIN 2016-75404-P AEI/FEDER.

• P. Bibiloni also benefited from the fellowship FPI/1645/2014 of

the Conselleria d’Educació, Cultura i Universitats of the Govern de les Illes Balears under an operational program co-financed by the European Social Fund.

7

Thank you

8