A novel filtering model for the erotic images embedded in web pages

Rapid development of various resources available on the Internet, the network just acts as a double-edged sword, which spreads useful and harmful information to the users simultaneously. In this paper, a novel filtering model for the erotic images embedded in web pages is proposed. It not only detects network eroticism, but also screens the web pages containing erotic images. For achieving a real-time processing in a dynamic network environment, the model is implemented by a BHO technique and its every part is designed on a cost-effectiveness analysis. The recognition accuracy of erotic images is improved significantly by introducing color correction into an image filter and adopting window mechanism into a text detector. Experimental results show that the proposed method obtains an excellent performance in terms of both efficiency and accuracy.