Abstract
THE NOVEL METHOD FOR RECOGNITION OF AMERICAN SIGN LANGUAGE WITH RING PROJECTION AND DISCRETE WAVELET TRANSFORM
Murat Taşkıran*, Sibel Çimen and Zehra Gülru Çam Taşkıran
ABSTRACT
Sign Language is a language that allows individuals with hearing or speech impairment to communicate with themselves and their surroundings and has the feature of not being a universal language. This language, which is not universal, is suitable for image processing as it is expressed by hand signals in general terms. In this study, firstly, we performed the skin color detection process to determine the position of the hand gesture, and then extract the edge regions of the hand gestures using the sobel filter and discrete two-dimensional wavelet transform. The obtained hand gesture image was used with the ring projection technique and the two-dimensional image was reduced to one-dimensional. The feature vectors were obtained by applying discrete wavelet transform to obtained the one-dimensional matrise. Classification of hand gestures was made with the Generalized Regression Neural Network (GRNN) using the obtained feature vectors. As a result of this study, 90.44% test accuracy was obtained.
[Full Text Article] [Download Certificate]