Sensitive Information Filtering Research Based On Contents

Nowadays it's the time of network and information, which the network, as its topic, assigns the era an informative hallmark. However, as the network offers us the convenience for cognizing the world, it also provides the spread of harmful information, such as pornographic, violence and other unhealthy illegal information, which disturbs our daily life badly. Hence, information security problems have become a big challenge for us.The multimedia sensitive information filtering technology is a crucial research subject in network information security, of which the vital part is image sensitive detection. The image sensitive detection in its essence is a segmentation problem, and is proposed to segment the sensitive regions from the other normal regions in an image, resulted in a binary map. A host of technologies are included in image sensitive detection, such as image segmentation, feature extraction, classifier design, body-part detection, skin detection, and so on.Firstly, we propose a novel Semantically Constrained Skin Detection(SCSD) method. In our method introduces the semantic constraint, which is based on the dependence between skin pixels and human body parts to refine the detection. Secondly, key frame extraction algorithm with unsupervised clustering is proposed to deal with videos. Thirdly, this paper predicts the sensitive regions by combining the global and local sensitive image region recognition algorithms. The experiment results on two datasets demonstrate the superiority of our method to other state-of-the-arts methods.