Head Gesture Analysis using Matrix Group Displacement Algorithm

,


INTRODUCTION
Head movement detection from an image sequences is a hard task due to variability in position, pose, size, orientation, color, expression, occlusion, resolution and lighting condition, hence, it is essential to develop efficient and adaptable algorithms to detect human faces. Detection of faces has been studied extensively over the past years, with various approaches. Gestures in general are expressive meaningful body motions with the intent to convey information or interact with the environment. Gesture recognition is the process by which gestures made by the user are acknowledged by an intelligence system (Villanueva and Cabeza, 2007;Srinivasa and Grossberg, 2008;Anishenko et al., 2008;Murphy-Chutorian and Trivedi, 2009;Doshi, 2009;Junker et al., 2008).
Understanding of human motion is especially important in fields such as biomechanics, robotics, human-computer interfaces, surveillance and computer animation. Based on the scope of each application area, various aspects of human motion have been studied from different perspectives. For example, motor control and learning theorists have studied how the central nervous system creates and updates internal representations of limb dynamics, to perform complex, programmed movements in dynamic environments (Lin et al., 2007;Eisenstein et al., 2008;Suk et al., 2010;Tonga et al., 2007;Wu and Trivedi, 2008;Markin and Prakash, 2006;Choi and Kim, 2008;Bohme et al., 2008). Sports analysts have studied the relationship between human physiology and movement efficiency. Ergonomics experts have modeled and employed complex relationships between human movement patterns, anthropometrics and ranges of motion to produce designs that better suit the human body. The fields of robotics and animation have developed algorithms and systems that mimic human motion and hence give us lifelike robots and humanoids.
In this study a novel procedure for static head-pose estimation and a new algorithm for head gesture position are presented. Visual tracking is integrated into the novel MGDA system for measuring the position and orientation of a head. This system consists of interconnected modules that detect head position, provide initial estimates of the head's pose and can be implemented to continuously track head position and orientation.

MATERIALS AND METHODS
MGDA image-based gesture recognition system is used. The system comprises three main parts: • Image processing and extraction of important clues, such as face or head position • Tracking gesture features, such as face poses • Gesture interpretation based on predefined gestures Figure 1 illustrates the operation of MGDA. Figure 2 shows images for practical (imperfect) head movements covering the five basic moves namely; left, right, straight, up, down.
Up Down Straight 2 Essentially, there are two basic movement sets: Even though the basic gesture sets are considered independent of each other; movements within each set must pass through the straight position. Figure 9 illustrate the two independent sets.
It is clear from Fig. 9 that there are two planes, one for [Up→Down] and the other for [Left→Right]. However, the movement translation should pass through centre (Straight) for standard movements or through a displacement of the centre point.
To move in between the two planes, a straight point position or its derivative is a requirement. The use of sequence numbers can be a substitute to discrete time intervals; hence, a transitional connection can be made between the two recognized planes as shown in Fig. 10.
The presence of two parallel planes describing Up→Down and Left→Right transitions leads to the following deductions: • Between Planes:

CONCLUSION
The presented Novel head pose scheme using MGDA that produces motion sequences is a feasible strategy for the identification of head poses and reconstruction of a sample continuous signal stream. It is demonstrated that such approach works well for static head images with non-ideal poses. The developed algorithm simplifies the process of acceptance rejection of non-relevant gestures.