Face

Biometric is area of technology, whose usage is still increasing - it can automate the process of human identification based on physical or behavioral characteristics. As a result of the computer performance evolution, new techniques for user identity recognition and verification has appeared based on the biometric features that are unique to each human.

Many biometric techniques have been developed in recent years which use selected human characteristics for recognition and identification. These include fingerprint, iris, hand geometry, retina, voice and facial recognition. Each biometric method has its strengths and weakness, and the choice of the specific method depends upon the requirements of the application. Among them, face recognition is one of the most frequently used to discriminate authorized and unauthorized persons.

Face recognition has the advantage of ubiquity and of being universal over other major biometrics, in that everyone has face and everyone readily displays the face. Uniqueness, another desirable characteristic for a biometric, is hard to claim at current levels of accuracy. Since face shape, especially when young, is heavily influenced genotype, identical twins are very hard to tell apart with this technology.

Face recognition systems have wide spectrum of applications. One of the most important is the access control system. If the face of the person who is trying to get access matches the pattern stored in the database, that person shall be granted with appropriate access rights. It is also often required to make verification of the person’s identity. For example – camera with the face recognition system can be installed in cash machine to check whether the person owning the card and trying to withdraw money is really the card owner, not a card thief. And the most important of the face recognition application is a security. With proper configuration of cameras it is possible to acquire face images without active participation of the subject. It is much needed in surveillance systems which rely on passive acquisition by capturing the face image without the knowledge of the person being imaged (for example airports to avoid the terrorist threat).

Methods used in the face recognition can be classified into:

  • Image feature based
  • Geometry feature based

First type methods try to estimate the correlation between a face and one or more templates. The result is later used during recognition. What is important, this kind of methods capture and analyze the global features of the face. Successful and efficient templates can be constructed using tools like:

  • Principal Component Analysis (PCA) (most frequently used in recognition)
  • Independent Component Analysis (ICA)
  • Fisher’s Linear Discriminate (FLD)
  • Kernel Methods
  • Support Vector Machines (SVM)

Geometry features based methods works quite different. They concentrate on local facial features and their geometrical relationships. For example, these geometrical relationships (metrics) can be used:

  • Inter-eye distance
  • distance between the lips and the nose
  • distance between the nose tip and the eyes
  • distance between the lips and the line joining the two eyes
  • shape of the face
  • ratio of the dimensions of the bounding box of the face
  • width of the lips

One of the first algorithm used in geometry feature based method was Elastic Bunch Graph Matching (EBGM), where local facial features were based on the Gabor wavelet transform.

Nowadays, there is new approach to the face recognition problem to use not only flat (2D) face image, but also the spatial data from the 3D scanner. There are many advantages in 3D face recognition compared to 2D method. First of all 2D images are very sensitive to illumination changes. The light collected from a face is a function of the geometry of the face, properties of the light source and the properties of the camera. In 3D images, variations in illumination only affect the texture of the face, yet the captured facial shape remains intact. Another difference in 2D and 3D face recognition is the effect of pose variation. In 2D images the angle of the face on image is very important, the higher the angle, the less information for the recognize process we have. In 3D images the face is scanned in very wide angle, so the pose of the face can be different and does not introduce an error to the recognition process. Also traditional 2D image-based face recognition focuses on high-contrast areas of the face such as eyes, mouth, nose and face boundary because low contrast areas such as the jaw  and cheeks are difficult to describe from intensity images. 3D images, on the other hand, make no distinction between high- and low-contrast areas.