PROJECT TITLE :

Accelerometer-based hand gesture recognition by Neural network and similarity matching - 2016

ABSTRACT:

In this paper, we have a tendency to gift an accelerometer-primarily based pen-type sensing device and a user-independent hand gesture recognition algorithm. Users can hold the device to perform hand gestures with their preferred handheld designs. Gestures in our system are divided into 2 varieties: the basic gesture and therefore the advanced gesture, that will be represented as a basic gesture sequence. A dictionary of 24 gestures, as well as 8 basic gestures and sixteen complex gestures, is outlined. An effective segmentation algorithm is developed to spot individual basic gesture motion intervals automatically. Through segmentation, each complicated gesture is segmented into many basic gestures. Based on the kinematics characteristics of the fundamental gesture, twenty five options are extracted to train the feedforward neural network model. For basic gesture recognition, the input gestures are classified directly by the feedforward neural network classifier. Nevertheless, the input advanced gestures undergo an extra similarity matching procedure to identify the foremost similar sequences. The proposed recognition algorithm achieves nearly perfect user-dependent and user-freelance recognition accuracies for both basic and complicated gestures. Experimental results based mostly on 5 subjects, totaling 1600 trajectories, have successfully validated the effectiveness of the feedforward neural network and similarity matching-based gesture recognition algorithm.