Hand Gesture Recognition Based on LPQ and MRELBP Descriptors

Gesture is an important communication tool, either for human-to-human communication, particularly to the estimated 70 million hearing-impaired people across the globe or for Human-Machine communication such as the human-machine interaction (HMI). In this paper, we propose an implementation of a static and dynamic gesture recognition system using three databases (American Sign Language ASL, Arabic Sign Language ArSL, and dynamic Clic) in an uniform background. The system is based on the Local Phase Quantization (LPQ) and the Median Robust Extended Local Binary Pattern (MRELBP) where the recognition task is applied using the Radial Basis Function Neural Network (RBF-NN) and the Support Vector Machine (SVM) classification techniques. Satisfactory results were obtained where the best recognition rate reached 97,5 % for American Language using LPQ descriptor.