A Pornographic Video Detection Method Based on Semi-supervised Learning on Graphs

In the field of pornographic video detection, it is relatively difficult to acquire the complete foreground region in video shots. It is necessary for further research in order to study the difficulty comprehensively. Therefore, this paper proposes an identification method of pornographic video based on semi-supervised learning on graphs. The method first obtains key frames from each video shot, and computes the inter-frame difference between key frames and adjacent multi-frame images to acquire the moving foreground region. Secondly, considering the moving foreground region as prior information, it applies semi-supervised learning on graph method to extract a complete foreground region and obtain skin region. Finally, the proposed method detects pornographic video content based on the features of skin region. The experimental results show that this method can correctly detect pornographic video with a high accuracy (up to 87%) and it is thus appropriate for the pornographic video detection.