An Efficient Feature Extraction of Isolated Word for Dynamic Sign Language Classification

Abstract

In image processing, feature extraction acts a key role. It is very imperative to know and extract the required features for further assessment. In this paper, the feature extraction of Arabic isolated sign language word based on chain code model is proposed. The features are extracted from the hand trajectory tracking, features obtained of the single hand or two hands that enter to a classifier which can determine the meaning of the gesture. In this study, More than forty isolated sign words are collected in collaboration with the Iraqi Ministry of Labor and Social Affairs. Four isolated words were taken as an example. The features were extracted from the isolated words; these features represent the feature vector of the isolated word that is used in the classification stage.