Which algorithm is used in face recognition

Digital face recognition with just one photo

"A single photo can be used to train the system," says Hung-Son Le, describing the peculiarity of its development in an interview with pressetext.

The basic handling of the photos is based on the so-called "Hidden Markov Model" (HMM). "HMM is a statistical tool that can be used in many areas to model patterns," explains Hung-Son Le. The method is widespread and is already in use for face recognition. "Previous access to the HMM required a lot of training photos," says the researcher, explaining the main difference to his model. This can recognize a face with only one photo for reference.

"I use another method to improve contrast and visual details," says Hung-Son Le, an additional strength of his set of algorithms. It can thus deal with insufficiently or excessively exposed images or different facial expressions. Sunglasses and especially disguises remain problematic, as with all face recognition algorithms.

The system was tested using international standards such as the Face Recognition Technology (FERET) database and, according to Hung-Son Les, achieved better results than current solutions. Commercial applications of the research results are already under development, including a web search engine for faces.

Hung-Son Le claims that he is not connected to the Swedish start-up Polar Rose, whose face search is currently in the closed beta phase. "To the best of my knowledge, it doesn't work with automatic face recognition," he says of the Polar Rose search. That is also true, as the company confirms at pressetext's request. The current beta version can find faces in photos, but does not yet contain any algorithms for automatic comparison. A corresponding functionality is also being planned. (pte / hal)