A Robust and Novel Approach to Identify the Nudity of a Person

Abstract

The nude image is an important part in adult images spreading on the Internet. As the development of the Internet, children are easily exposed on the pornography through web browsers. To block adult images, content-based image retrieval technique is employed for nude image identification. In this paper a novel algorithm to detect explicit content or pornographic images is proposed using the transformation from the RGB model color to the YCbCr. First, the background is removed to obtain the rectangular region of interesting based on the detection of skinlike pixels. For each input image, color feature is used to retrieve 900 most similar images from the image database which contains both nude and non-nude images. However, if the retrieved images contains more than threshold nude pixels, then the input one is identified as a nude image. Otherwise, it is identified as a non-nude image. Experiment results have shown the effectiveness of the proposed method. Keywords: Explicit content, Gabor filter, Nude image identification, Skin-pixel, YCbCr.