Naeratuse detektor näo kontrollpunktide liikumise põhjal
Date
2014
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
Inimese ja arvuti suhtlus on kahtlemata tänapäeva ühiskonna väga tähtis osa. Et seda veelgi parandada on võimalik luua
süsteeme, kus arvuti reageerib inimese liigutustele või näoilmetele. Naeratamine on ilmselt näoilme, mis annab inimese
kohta kõige rohkem informatsiooni. Selles lõputöös kirjeldame algoritmi, mis suudab tuvastada seda, kui inimene naeratab.
Selleks leiame kõigepealt Viola-Jones'i algoritmi abil näo asukoha. Seejärel leiame vajalikele näoosadele vastavad
kontrollpunktid ning jälgime nende liikumist järgmiste videokaadrite jooksul. Tuvastatud liikumise järgi otsustab algoritm, kas inimene naeratab või mitte.
Human and computer interaction is without doubt a really important part of our modern society. In order to improve it even further it is possible to develop computer systems that react to gestures or facial expressions of its user. Smiling is an expression that gives probably the most information about a person. In this thesis we describe an algorithm that understands when a person is smiling. To achieve that we first detect a face of a person using the Viola-Jones algorithm. After that several facial reference points are located and then tracked across several consequent frames using optical flow. The motion of these points is analyzed and the face is classified as smiling or not smiling.
Human and computer interaction is without doubt a really important part of our modern society. In order to improve it even further it is possible to develop computer systems that react to gestures or facial expressions of its user. Smiling is an expression that gives probably the most information about a person. In this thesis we describe an algorithm that understands when a person is smiling. To achieve that we first detect a face of a person using the Viola-Jones algorithm. After that several facial reference points are located and then tracked across several consequent frames using optical flow. The motion of these points is analyzed and the face is classified as smiling or not smiling.