Expert's recommendations
Face recognition is a security feature for identifying people. It is quite similar to fingerprint scanning and iris scanning. Initially face recognition was employed in government agencies for security purposes. These systems were popular in airports too. Even some companies used this technology to identify their employees and to prevent proxy-punching. This utility has found its way into laptops and notebooks as well. Digital cameras and camera mobile phones offer face detection to enable produce clear faces in photographs.
How does face recognition work?
A database of people’s faces is maintained by the system that handles face recognition. Whenever a face needs to be identified, a photograph of the person’s face is taken and compared to the faces present in the database, to see if a match is found.
There are usually 3 parts to a face recognition system – face detector, eye localiser and face recogniser. The face detector detects the face, eliminating any other detail, not related to the face (like the background). It identifies the facial region, leaving out the non-facial region in the photograph of the person to be identified. The eye-localiser finds the location of the eyes, so that the position of the face can be identified better. The recogniser, then checks the database to find a match.
Like any other technology, face recognition has also evolved and come a long way from what it was initially. The 2D face recognition was one of the earliest techniques used. It maintained details of people’s faces as seen 2 dimensionally. Details like width of the eye, width of the nose, jaw line, distance between the eyes, cheek bone shape and the like were used for comparison. This kind of face recognition was not very accurate. Difference in ambient lighting or a face that is not directly looking into the camera, or a change in facial expression did not produce expected results. Hence 2D face recognition can be safely used only in companies where employees can show their faces directly to the camera for identification. Even laptops and notebooks use 2D recognition. However, in airports or government agencies, where suspicious people have to be identified, even when they’re walking in a crowd, 2D technology might not work effectively.
Advancement in face recognition gave birth to the 3D identification technique. This stepped up technique, uses features like contours of the eye sockets, nose, chin, peaks and valleys on the face for identification. The database stores such details of faces as well. The advantage of 3D over 2D method is that 3D face detection works well even if the face is turned at 900 to the camera. Also, it is independent of lighting environment and facial expressions. When a suspicious person in a crowd needs to be identified, an image of that person is clicked with a camera. The face recognition system then identifies the shape of the face, determines the curves, the peaks, the valleys and other such distinguishable factors. These details are converted into a set of patterns with codes given to each of them. These patterns with codes are compared with the information in the database to find a match.
A more advanced method is Surface Texture Analysis (STA). This technique does not scan the entire face but a patch of skin on it. This patch is divided into blocks. The skin texture, the pores on the skin and other such characteristics are converted to a code, which is used for comparison.
Many face recognition systems include a combination of the 2D, 3D and STA for obtaining accurate identification of faces. Among the many biometric recognition systems, face recognition is the most non-intrusive method. Hence it finds wide application in highly secure areas. The big leaps and bounds that technology has taken has made face recognition very popular and accurate as well.
Related posts:
