Recognition Systems

Recognition Systems
the following steps. 1. Identify and distinguish these characteristics may be at the roundness of the face, eye position, etc. The roundness can be determined by the uniformity of the curvature of the contour of the face. A larger number would round a round face. The eye position can be the vertical distance from the chin to the pupil. 2. Label map and the numerical values to face the person may have a roundness of 2 (10) and their eyes may be 8 inches above the tip of the chin. The value of (2.8) can be stored in memory with the name of the person associated with it. 3. Determine the importance of each feature This is the hard part. The face of one person can match "sufficiently close" several names in the database based on the traits measured. On the contrary, several people also may coincide with a particular point of the stored data. For example, two people with traits (2.1,8) and (2,8.1) can coincide with (2.8) Stored data point above. Without But some features are more important because they are more useful for distinguishing faces. To determine the importance of each trait is, the team must "learn" through neural network logic. After much trial and error (ie, the "learning process" that occurs as part of the development phase), the algorithm can determine that the roundness is more important, and therefore the (2,8.1) person was identified as the party of truth.