gb2s
-->

Face

face
 What is face recognition?

 

During the last decades, several different techniques have been proposed for computer recognition of human faces.

Facial feature extraction methods can be classified into two main categories, according to the way of obtaining the face patterns:

  • “Holistic approach” (Template matching) which considers the hole face region in an image as the system’s input data and represents each face as a vector whose components codify the grey level of each face pixel.

  • “Feature approach” (Geometric, feature-based matching) which establishes certain facial landmarks related to face elements, such as eyes, nose, mouth and ears and computes features as distances between landmarks, relative positions or elements' sizes.

A big majority of these systems are based on the holistic approach because the template approach is more reliable than the feature one, and its implementation is simpler.

 

 How does face analysis work?

 

Face identification is carried out in three steps:

  1. Face detection in which a face is located within an image. All the images in the database are converted to gray scale and normalized by using Histogram Equalization. After detecting a face within an image, the face area is cut and both the background and the face surroundings are suppressed to avoid extracting features of non faces. Then the face-region image is scaled to 92 × 112 pixels, in order to compare color intensity between related pixels.

  2. Pattern construction where normalization and face feature extraction take place. All the images in the testing set are classified with classifiers like the k-Nearest Neighbors (kNN) method, one of the most popular distance-based classifying algorithms. Then, for each method, a correct identification rate is computed.

  3. Pattern recognition in which the previously located face is linked to an individual, who has already been enrolled in the system. On each verification attempt, the presented image’s template (sample) is compared with the stored user template (pattern). The verification process is considered as “successful” by the system if the difference between these templates is smaller than a prefixed threshold.

 

 Applications

 

  • Mobile security system
  • Airport Security
  • Age restriction for social networks

 

  Our Projects

 

 

Category: /

Grup of Biometrics, Biosignals and Security (GB2S)