Patent for computerized image-annotation system issued to Li and Wang

Jia Li, associate professor of statistics at Penn State, and James Z. Wang, professor of information sciences and technology at Penn State, have been awarded the United States patent titled "Real-Time Computerized Annotation of Pictures." Li and Wang developed a unique computer system called ALIPR (Automatic Linguistic Indexing of Pictures in Real-Time) that can suggest, in real time, several words describing the content of any general-purpose photograph using the pixel information alone. The automatic image-tagging system uses novel methods in statistical learning and data mining to link pictorial characteristics with English words.

The ALIPR technology can be used to enhance image search engines and to facilitate methods used by computers to manage large-scale collections of pictures. Members of the public are invited to help improve ALIPR's accuracy by visiting http://www.alipr.com, where they can upload photographs and evaluate whether the keywords that ALIPR uses to describe the photographs are appropriate. A more detailed story about the development of the ALIPR software is online at http://www.science.psu.edu/news-and-events/2008-news/Li10-2008.htm.

Earlier this year, Li and Wang were issued another patent as a result of their research into improving computer security. This earlier patent is for a CAPTCHA (Completely Automated Public Test to Tell Computers and Humans Apart) image-generation system, which is shown to be more effective than computer-security systems that use distorted texts to determine if a real user or a computer robotic program is trying to gain access to a website. The new system, which Li and Wang developed in collaboration with Ritendra Datta, who was a Penn State graduate student at the time of the research, uses modified colors, shapes, and textures to create an image or a mosaic. Users must then identify the content of the image or determine the geometric center of any tile in the mosaic. The new CAPTCHA software helps to reduce the frequency of online hackers, and with further research and development, such technologies could one day prevent hacking altogether. The research leading to both patented systems was funded by the National Science Foundation.

Contacts: 
Last Updated July 25, 2011