01 - Introduction

About the project

• choose a topic and make a software on it
  • you’re allowed to get any software as long as you modify enough
  • not forks, or forks + some other features
  • you must know all the topics in the syllabus

Definition

• biometrics refers to the study and use of methods for detect and measure the characteristics of living organisms and draw comparatively classifications and laws
  • in CS: automatic identification or verification of the identity of a person based on physical or behavioral characteristics
    • identification: who are you
    • verification: what prove that you are who you declare
  • Biometric Consortium: automatic recognition of a person according to discriminative characteristics

What they does

• biometric system are special systems that do pattern recognition
  • pattern recognition is at the basis on many topics, like computer vision
  • pattern recognition is a special model that reads information that is necessary to summarize the characteristics of some objects (features)
    • when we have features we can measure the distance between two objects (or class of objects) and see if they are enough similar
  • problems:
    1. what is a good distance measure
       • quick, easy to calculate
       • we compute the difference between the two similarities
    2. which are the best features
       • must be able to distinguish between two objects belonging to the same class
       • they’re good only if the can distinguish
    3. what is the difference margin to accept
       • it is a margin of uncertainty to agree that two objects are similar

Classies and patterns

We distinguish between classies and pattern in several ways:

• content based image retrieval

  • classies: type of objects
  • pattern: any kind of features that allow to distinguish between a face and a flower

• Recognition

  • classies: subclasses of same type
  • pattern: allows to distinguis a subclass

• Biometrics

  • classies: individuals
  • pattern: allows to distinguish among individuals

Uniqueness of a person

• basic assumption: each person is unique

  • it’s true, but we all share a lot of similiarities (features)

• main issues

  • determine the unique features able to identify a person
  • find reliable techniques to measure such features
  • devise reliable algorithms to recognize/classify a person according to the measured features