Arabic Sign Language Recognition (ArSL) Approach Using Support Vector Machine
Mohamed Arafa Ali, Mohamed Rabie Ewis, Gamal Eid Mohamed, Heba Hamdy Ali, Hossam M. Moftah · 2017
Sign Language Recognition is an important method that improves communication between deaf people and normal people. In this paper, we introduce an approach that recognizes signs of alphabets of Arabic sign language. We worked on 28 signs of distinctive signs on the part of alphabetical characters of Arabic language; our experiments have applied on the dataset that we prepared on standard Arabic Sign Language. The dataset consists of 1400 images, 50 images per each category and divided into training 80% and testing 20%, the proposed approach extracts hand from the captured frame of live video, then performs preprocessing and detection operations on a sign. We used Dense SIFT technique to extract feature vectors that represent the image. In the classification phase, we used both multi class Support Vector Machine and Logistic Regression methods, the results of our experiments achieved up to 96% using Support Vector Machine.