WIPE :: Wavelet Image Pornography Elimination

With freedom of speech comes the consequence that people may display objectionable information, such as pornographic images, online. In 1997, James Wang developed the Wavelet Image Pornography Elimination (WIPE) filter, the first real-time image-based filter that can tell the level of indication that an image is objectionable with high accuracy suitable for real-world deployment. The WIPE system was subsequently licensed by Stanford University and related technologies have been used in large online ISPs. The filtering process is based on the image's pixels. Filters such as this are important to protect internet users from offensive content. Simple mistakes by internet users, such as typing in 'male' instead of 'mail', can result in objectionable websites and images to appear. Uses may include blocking of objectionable material for children.

Project Director

The project has been conducted by James Z. Wang at The Pennsylvania State University.

Project Background

This content-based automatic objectionable image detection engine was developed by J. Z. Wang in 1997. It was the earlist automated learning-based system for pornographic image filter with high accuracy both for classifying porn images and for classifying benign images. The research work was conducted at Stanford University with G. Wiederhold. The program does not consider any information other than the pixels of the image itself. The current training database is relatively small. Configurations of the algorithm may be changed without notice. The server has been under attack and it can be down some times. But you can read the publications if it is not working properly.

Demonstration

   We have not been maintaining this demo. But a version of it is online. We may update the demo in the future.

Related Publications

1. James Z. Wang, Jia Li, Gio Wiederhold and Oscar Firschein, ``System for Screening Objectionable Images,'' Computer Communications, vol. 21, no. 15, pp. 1355-1360, Elsevier, 1998. (download) (citations)
2. James Z. Wang, Jia Li, Gio Wiederhold and Oscar Firschein, ``Classifying Objectionable Websites Based on Image Content,'' Lecture Notes in Computer Science, Special issue on interactive distributed multimedia systems and telecommunication services, Oslo, Norway, Thomas Plagemann and Vera Goebel (eds.), vol. 1483, pp. 113-124, Springer-Verlag, September 1998. (download) (citations)
3. James Z. Wang, Gio Wiederhold and Oscar Firschein, ``System for Screening Objectionable Images Using Daubechies' Wavelets and Color Histograms,'' Lecture Notes in Computer Science, Special issue on interactive distributed multimedia systems and telecommunication services, Darmstadt, Germany, Ralf Steinmetz and Lars C. Wolf (eds.), vol. 1309, pp. 20-30, Springer-Verlag, September 1997. (download) (citations)

More Information

1. Research Group
2. Related Media Annotation projects done by this group
3. Similar technology is being used in some of our biomedical projects