A Generic Approach to Naked Images Detection

J.-S. Lee and S.-C. Su (Taiwan)

Keywords

Naked image detection, eigennipple, nipple space

Abstract

Because of the popularization of the Internet access, governments and parents have been facing a serious problem that juveniles can intentionally or unintentionally access naked images from the Internet in effortless ways. How to effectively block or filter out pornography has become an important issue. In this paper, we propose a generic naked image detection algorithm. Contrast to the existing methods relying on chromatic information our algorithm detects naked images by only using intensity information. This way makes our algorithm inherently possess immunity against skin color deviation coming from special lighting that is a thorny problem to the existing methods. In addition, this also enables our algorithm to handle gray-tone naked images that is beyond traditional methods’ capability. We make decision based on examining if the role exposes her sensitive areas, i.e. nipples, instead of the skin area size. Such that our algorithm can achieve more convincing results.

Important Links:



Go Back