Static Persian Sign Language Recognition usingKernel-based Feature Extraction

The most effective way for deaf peoplecommunication is sign language. Since most people are notfamiliar with this language, there is a requirement for a signlanguage translator system. This would be a useful toolspecifically in emergency situations. A further need is facilitationof deaf people communication in cyberspace. Sign languagegestures can be divided in two groups, including gesturesrepresent the alphabets and those which are arbitrary signsrepresenting specific concepts. The first group is usuallyintroduced by the pose of hands and they are called postureswhile the second group usually includes motion of the hands. Thispaper evaluates the efficiency of kernel based feature extractionmethods including kernel principle component analysis (KPCA)and kernel discriminant analysis (KDA) on Persian sign language(PSL) postures. To compare the impact of features on signs’recognition rate, classifiers such as minimum distance, supportvector machine (SVM) and Neural network (NN) is used.Experimental trials indicate higher recognition rate for thekernel-based methods in comparison to those of other techniquesand also previous works on PSL recognition.