期刊名称：International Journal of Advanced Research in Computer Engineering & Technology (IJARCET)
印刷版ISSN：2278-1323
出版年度：2017
卷号：6
期号：11
页码：1776-1781
出版社：Shri Pannalal Research Institute of Technolgy
摘要：The occurrence of malignant melanoma has beenincreasing worldwide. A well-organized non-invasive computeraideddiagnosis (CAD) is seen as a solution to makeidentification process faster, and accessible to a largepopulation. Such automated system relies on three things:reliable lesion segmentation, pertinent features’ extraction andgood lesion classifier. Automated melanoma recognition indermoscopy images is a very challenging task due to the lowcontrast of skin lesions, the huge intraclass variation ofmelanomas, the high degree of visual similarity betweenmelanoma and non-melanoma lesions, and the existence ofmany artifacts in the image.
