Nipple Detection for Obscene Pictures

This paper presents a research on object detection using image processing and neural network, entitled Nipple Detection for Obscene Pictures. Our research work is designed to enhance the effectiveness of pornography detection system by locating nipples in the picture to make sure that the given picture really contains human nudity. Our nipple detection model composing of human skin detection which is used to remove non-skin area and nipple detection which is used to detect nipple on human skin area. Our approach is shown to be effective for a wide range of shades and colors of skin and human configurations. It is validated for locating nipple existence for obscene pictures. In terms of scalability, our approach can be modified to support distributed processing. The system presents excellent performance on a test set of 130 uncontrolled obscene pictures, and a test set of 500 assorted control benign pictures drawn from a wide variety of sources. Key-Words: Object Detection, Computer Vision, Erotica/Pornography, Content Based Retrieval.