Abstract
This paper describes a recognition system that is designed to aid users of American Sign Language. American Sign Language is growing in popularity, and such systems can be used to help improve the communication barrier between ASL sign language users and people who don't understand the language. The proposed preliminary system is designed to detect ten words in the ASL language. Sensors positioned on both wrists are used to collect data that will reflect the motions of the hands. The data is processed and trained using the Support Vector Machine, and Discriminant analysis classifiers using a selective combination of features were explored. Data was collected with Inertial Measurement Unit sensors strapped on the index and middle fingers to simulate a wearable ring. This data, combined with the data collected from the wrists, are trained using the Discriminant analysis classifier, and the system achieved a classification accuracy of 82%.