Hand gesture recognition using EMD and VMD techniques

A new approach based on decomposition techniques for better feature extraction of recognition of dynamic hand gesture recognition system. In this paper we are analyzing a comparison of two useful noise removal techniques, empirical mode decomposition (EMD) and variation mode decomposition (VMD) for strong occlusions, nonstationary and weak robustness complex backgrounds. So implemented results show the feature extraction by using EMD with different values of intrinsic mode function (IMFs) and VMD with different values of modes and obtain a noise free signal. A non-stationary electromyography (EMG) signal of hand movement is measured of VIVA (Vision for Intelligent Vehicles and Applications) dataset, where eight subjects are performing 19 types of dynamic hand gestures in a vehicle and this is captured by Microsoft kinetic.