Developing Classifier Model for Hand Gesture Recognition for Application of Human Computer Interaction (HCI)

Abstract

The hand gesture technique that is regarded as the natural and easy method for the human-machine interaction, has paved way for the development of the multitudes of applications. The hand gestures basically employed in most of the application are either sensor based or the vision based. In case of verbal communication the gesture depiction involves the application of the natural and the bare hand gestures. So the paper proposes a bare hand gesture recognition with the light in variance conditions, involving the image cropping algorithm in the preprocessing, considering only the region of interest. The mapping of the image oriented histogram is primarily done utilizing the Euclidean distance method and further supervised neural network are trained using the images mapped, to have a better recognition of images with the same gestures under different light intensities.