Hand Gesture Recognition Through Depth Sensors

Introduction

The gesture is a type of nonverbal communication, and it is the oldest form of communication between humans. The gesture involves the use of several body parts, primarily the hand and face. Gestures were used by ancient men to express information such as food/prey for hunting, supply of water, knowledge about their enemy, a plea for help, and so on. Gestures are bodily actions that communicate information. If we will not consider the world of computer study about the contact between humans for a while, we may readily see that our communication uses a wide range of gestures. Hand gesture has become a fantastic option for expressing simple concepts, interpreted by the system of gesture recognition, and transformed into events. The possibility of interacting with computer systems to provide meaningful information for the interaction with human-computer is presented in various hand movements classified on the basis of the distinctive shapes of hand, finger, or orientation. Advanced technologies require that varied hand gestures and their recognition, classification, and interpretation so they can be applied in a wide range of computing applications.

The process of automatically recognizing and analyzing human behaviors from data is known as human action recognition, which can be acquired from various types of sensors such as RGB cameras, depth cameras, range sensors, wearable inertial sensors, or other modality-type sensors. Nowadays gestures are widely used for many applications in various fields. As a novel input modality in human-computer interaction, gesture recognition systems enable a more intuitive and convenient means of interacting. In human-computer interaction, hand gesture recognition provides the fewest limitations for the user (HCI). Hand gestures have a wide variety of applications in Human-Computer Interaction (HCI), which can ensure speed of communication and provide an easier user-friendly and aesthetic environment, provide non-professional contact with the computer from a distance for comfort and security, and control complex and virtual environments in a much easier approach (Andrea, 2014). In contrast, the application of hand gestures requires the user to be an expert and well-trained to utilize and grasp the meaning of various gestures (Samata & Kinage, 2015). A group of manual gestures can be employed for each application to complete its functions. From the point of the computer of view, the external factors, such as light, skin color, location, and orientation of the hand affect the performance of the recognition of hand motions (Vaibhavi et al., 2014). A classification of gesture recognition techniques has been shown in figure 1.

Figure 1.

Classification of gesture recognition

Background

In the research field, hand gesture recognition has been introduced for 38 years (Prashan, 2014). A hand glove was proposed by Zimmerman in 1977 that was able to detect the number of fingers bending (Praveen & Shreya, 2015). In 1983 Gary Grimes has developed a system to determine if a thumb or finger touches any part of the hand (Laura, Angelo & Paolo, 2008). The word gesture recognition refers collectively to an entire process of recording human motions into semantic-significant orders for their representation and conversion. Human-computer interaction is dependent on two basic technologies which are named contact-based and vision-based. The physical user interaction with the interfacing device is a foundation of contact-based devices used for hand gesture detection systems. The interfacing devices can be assembled with various devices such as multi-touch screens, data gloves, and accelerometers, etc. Contact-based devices for hand gesture recognition can be classified into five types such as mechanical, haptics, ultrasonic, inertial, and magnetic (Kanniche, 2009). In the research work of Chen et.al (2013) an evaluation of articulated 3D body modelling has been represented for estimating the human position and recognizing human motion from images in depth.