M.Sc. Tezi Görüntüleme | |||||||||||||||||||||
|
|
||||||||||||||||||||
Summary: In this thesis, it is aimed to interpret sign language used by deaf people based on computer. For this, it is required to determine representations for words used in the sign language and perform classification by using these representations. First of all, hand motion tracking in the video where datas are extracted from was implemented. For object tracking, mean shift algorithm was used. Thus, change in hand shape in time and trajectory of hand were obtained.In the next step, the feature vectors were extracted from the obtained trajectory. To extract hand shape feature vectors, Zernike moments were computed. Finally, hidden Markov model classification was proposed for various types of hand gesture recognition. The decision regarding in recognition was made based on similarity among signs in the trained system.
Keywords: Sign Language, Mean-Shift Algorithm, Zernike Moments, Hidden Markov Model
|