Hand and Wrist Localization Approach for Features Extraction in Arabic Sign Language Recognition

This paper proposes a new hand detection and wrist localization method which presents an important step in the hand gesture recognizing process. The wrist localization step has not been given much attention and the existing works are limited and include many conditions. Our proposed approach was evaluated on a public dataset whose obtained results underscore its performance. We highlight through a comparative study with existing work, the superiority of our approach and the importance of the wrist localization step. We also propose to benefit from our proposed method which can be applied in the sign language recognition domain, and more precisely in the Arabic digit sign language recognition.