It is a popular internet behavior that users browse through the pornographic websites for the purpose of satisfying their curiosity. Most users start to be image provider due to the anonymity, availability, immediacy, and interaction of internet. Pornographic images can be posted on almost every website because people tend to share whatever files they have with other users along with the offer of free homepage space service. The chart website is a typical example, which is a space with no payment mechanism and censorship. Almost all kinds of pornographic images are available on the website. It is a must to control and prevent such behavior.
In this thesis, we propose a nude image detection mechanism based on the techniques of skin color detection, face detection, and skin region detection. By utilizing internet packet, we can retrieve the user’s URL and then analyze whether the contents contain any inappropriate image. Finally, black and white URL lists will be produced to prevent the act.