Different Model for Hand Gesture Recognition with a Novel Line Feature Extraction

Hand gestures are commonly used for communication between both impaired community and normal people. Sign languages stand for the human languages of deaf people. They form the most growing domain of research worldwide. A number of techniques was developed in this area lately. It is recognized by means of deducing the features involved in the use of hand gesture. As a matter of fact, various approaches, namely the vision-based, the data-glove-based, the colored-marker and the Electromyogram (EMG) approaches have been utilized by researchers to recognize the different hand gestures implemented in many different fields such as the whole approaches which can be divided into four main categories, viz. Data Collection, Image Processing, Feature Extraction and Gesture Recognition. Only few of those categories have been discussed in this paper to be compared between the accuracy rates by applying Artificial Neural Network (ANN) classification. This classification is based on different models and a novel method for Real-Time Hand Gesture Recognition System (RTHGRS). The latter has used one line (fifty features) extracted from black and white processed images to recognize the numbers from (1-10) in Kurdish Sign Language (KurdSL) using one hand only with accuracy 98%.