Pencil in a pair of dots, put a curved line below and draw a circle around the whole lot. What have you got?

A face, of course.

But as simple as it seems, recognizing a face is no mean feat.

	Tony Tye, Post-Gazette Here's how a computer progam "sees" the faces of Henry Schneiderman, left, of the Carnegie Mellon Robotics Institute, and his assistant Gary Ackley. Schneiderman developed this computer vision device that can find human faces in a black-and-white image. The software puts a box around any face that it finds, then analyzes the details. Click photo for larger image.

Neuropsychologists debate whether people have an inborn ability to recognize faces, or whether it is a skill that develops from earliest infancy. It is a task of such difficulty and importance, however, that the brain has one area that is largely devoted just to faces.

And computer scientists, attempting to design machines that can pick out objects in images, have found that faces are a major challenge.

"Scientifically, it's interesting because faces have a great deal of variation from person to person," said Henry Schneiderman, a computer vision researcher at Carnegie Mellon University's Robotics Institute. Though they all have eyes, noses and mouths, their shapes and sizes vary widely, as does skin color and facial hair.

Machines also can be confused by lighting variation, sunglasses and other apparel. The orientation of the head -- frontal vs. side views -- can dramatically alter the ability of the computer to recognize a face as a face, said Schneiderman, who has developed the most accurate program in existence for detecting faces in still images and video.

Tony Tye, Post-Gazette Here's how a human photographer -- and his digital camera -- sees Schneiderman and Ackley. Click photo for larger image.

Though face detection is easy for most people, some suffer a perplexing disorder called face blindness, or prosopagnosia, which is an inability to discern the differences between faces, said Marlene Behrmann, a CMU professor of psychology who specializes in human perception.

One such sufferer, who is part of a research study by Behrmann, can't recognize his own children when he picks them up from day care. He relies on the day care workers to bring his children to him; failing that, he carries a "cheat sheet" of photographs that can help him make out who's who.

People with face blindness can use other clues, such as the sound of a voice, height or clothing to identify people. But recognizing the faces of people is a key skill for social animals such as humans; without it, people are profoundly impaired.

Face blindness can have dramatic effects on behavior, Behrmann noted. "One woman [in the study] is known as the friendliest person on the block," she said. Not readily knowing who is who when she encounters people on the street, she treats everybody as if she knows them. But other face blind people adopt the opposite behavior, withdrawing from society because of the constant strain of identifying people.

Elusive causes

It's not clear what causes this problem.

Post-Gazette files Try your hand at facial recognition. The composite face at left was made from the features of two prominent Republicans. The image at left is from two well-known entertainers who speak English with an accent. (Answers at end of story.) If you enjoy this little game, Web surfers can submit images to test, or possibly stump, Schneiderman's Face Detector at vasc.ri.cmu.edu/cgi-bin/demos/findface.cgi. Click photo for larger image.

Like everyone else, face-blind people have a specific region of the brain's temporal lobe that is activated when the person sees a face. In brain imaging studies, this area actually appears to be a bit more active in face-blind people than others, Behrmann said. The region is working harder in them, most likely, because it is struggling at the task.

In addition to the defined face region, other areas of the brain aid in identifying faces, she noted. The cause of face blindness may actually lie in one or more of these areas, or perhaps in the connections between these brain areas. She said she is now performing studies that examine these white-matter links in face-blind people.

It's possible there may be a genetic susceptibility to face blindness, Behrmann said. Also, people with face blindness may be susceptible to a reading problem known as developmental dyslexia.

But perceiving faces isn't easy even for people who don't suffer face blindness. "Faces are incredibly difficult," Behrmann said, noting that the brain "recruits" a good bit of its visual cortex to the task of face recognition.

Some neuropsychologists argue that babies are born with the ability to recognize faces, but that's not yet clear, she said. It's likely, however, that faces are one of the first things that babies recognize, whether that's an inborn ability or a learned skill.

"Remember that a baby's vision is not very good," Behrmann said. Faces may be one of the first things that come within eye shot of a baby, considering how close mothers and others tend to hold babies.

Though one brain area is activated by faces, it likely isn't used exclusively for faces. Other areas of the brain are activated for identifying other sorts of objects, such as houses or cars. Depending on a person's habits or interests, more or less of the brain may be recruited for specific tasks; a bird watcher, for instance, may gradually increase the amount of brain area that is activated for identifying birds.

Getting computers to 'see'

In the field of computer vision, faces pose a different sort of problem. Though face-blind people can detect faces, they have trouble telling the difference between them. Schneiderman said computers have less trouble telling the difference between faces than they do simply picking out faces from other objects in an image.

In developing a face detection program, Schneiderman and other computer vision researchers, such as former Robotics Institute director Takeo Kanade, can't tell the computer precisely what a face is supposed to look like. So part of the development process involves showing the computer examples of faces and non-faces and letting the computer program gradually develop its own statistical rules for determining what constitutes a face.

No one knows how the human brain represents images, but computers use numbers, with each number representing one point, or pixel, in an image. In black and white images, the larger the number, the brighter the pixel.

Schneiderman's face detector uses low-resolution black and white images measuring 24 by 32 pixels, or a total of 768 pixels. That means the computer has to analyze 768 numbers for patterns it thinks are faces.

Statistical analysis of what these computer programs have learned suggests that symmetry across the face, particularly the spacing of the eyes, is a critical element. The forehead, which has even skin texture and usually a high brightness level, also seems to be important, Schneiderman said.

Behavioral studies suggest that the configuration of eyes, nose and mouth also is important for face detection by humans, Behrmann said. When faces are turned upside down, people have trouble recognizing them. Likewise, changing the spacing of the eyes can have significant effects on the ability to detect a face. Changing hair or the spacing of ears has less effect.

Using a standard benchmark test for computer detection of faces, Schneiderman's program is capable of detecting 93 percent of the faces in a set of images while falsely identifying four objects as faces. Detecting a higher percentage of faces results in a larger number of objects falsely identified as faces.

Faces viewed from the side are harder to detect. "Somehow, an eye doesn't look as distinctive when you see it from the side," he said.

Schneiderman's Face Detector has been exhibited at the Science Museum of Minnesota and next week will be one of the technologies featured at Wired magazine's NextFest exhibition in San Francisco.

The Face Detector is being exhibited as a security technology; presumably it might be used to detect people who are in secure areas, or to pick out faces for identification in crowds.

But Schneiderman noted its first use was in photo processing. A company that builds equipment used in one-hour-type photo processing shops licensed the technology to locate eyes in photographs; the company used this as part of a system for automatically reducing "red eye" that can show up in some flash snapshots.

In addition to detecting faces, the program also can be applied to other types of objects. Eventually, Schneiderman envisions it being used to organize and search images produced by digital cameras.

"I think digital photography is going to change the way people take photographs," he explained. In his own case, he already has amassed thousands of images on his computer of his two children. If a computer could identify objects on its own, it should be possible to pick out photos of, say, a birthday party by searching for particular faces and birthday hats or balloons.