Page 159 - 2024-Vol20-Issue2

155 |
interaction with virtual objects, virtual assembly, and remote
operation. Communicative gestures are crucial in human in-
teraction and psychological research. Techniques based on
vision capture motion techniques can aid this research [4].
Gestures can be categorized into static gestures based on
hand shape and dynamic gestures influenced by hand move-
ments [5]. In ref. [6], Pinto et al defined hand posture as a
combination of position, orientation, and flexion observed at
a specific time. Static hand gestures involve a single posture
over time, allowing for interpretation through one or more
hand images. Simple signs for this gesture include ”OK” or
”STOP.”
In a video signal, a dynamic hand gesture [7] is a series of pos- A. Vision-Based Approaches
tures connected by motion over a brief period. These gestures
require identifying temporal context information and can be
static or dynamic in certain situations. Sign language is a com-
munication method using gestures and postures to communi-
cate with deaf or dumb people. These gestures can be used
in systems control, Augmented Reality, Gaming, robotics,
and vision-based applications. Via imaging devices, a sign
language recognition system interprets sign language into
corresponding text. However, there are complexities in this
process, as spoken languages change across different coun-
tries and regions, affecting the corresponding gestures [8].
Previous research shows that the deaf and dumb community
will increase by 2050 compared to what it currently exists
due to noise and other reasons, and with the increase in the
number of the deaf and dumb community compared to the
ignorance of ordinary people in sign language and the need of
the two communities to communicate, recognition of sign lan-
guage has become of great importance by using technological
methods [9].
II. HAND GESTURE MODELING
Modeling the hand is crucial for understanding posture and
gestures as interfaces in Human-Computer Interaction (HCI)
stages [10]. Hand modeling relies on kinematic structure for
accurate techniques [11]. Gestures can be modeled spatially
or temporally, focusing on posture characteristics in HCI ap-
plication environments and dynamic hand gestures in time
modeling. Spatial modeling for hand modeling in two- and
three-dimensional spaces [12, 13].
Fig.1 displays four 2D shape types: shape, motion, colored
marker, and deformable templates [12]. Geometric models are
based on fingertips and palm features, while non-geometric
models use features like silhouette, texture, color, contour,
edges, image moments, and eigenvectors.
Deformable templates or flexible models provide a flexible
level of the object shape change [14] to allow for a slight
change in the hand’s shape. The motion-based model can be
implemented concerning color cues for hand tracing, color
Murad & Alasadi
Fig. 1. Hand gesture modeling [16].
markers, or track and finger position estimation for hand shape
modeling [15]. The 3D model will be discussed in appearance-
based approaches in the section 3D model-based approach.
III. HAND GESTURE ANALYSIS APPROACHES
Identifying hand gestures involves obtaining data for specific
tasks using Vision-Based Data-Gloves and Color-Marker tech-
niques.
Vision-based approaches use cameras or more to obtain hu-
man motion, allowing devices to interpret gesture properties
like color and texture [17, 18]. However, challenges like light-
ing diversity, complex backgrounds, and clutter can arise, as
well as time, speed, durability, and computation efficiency.
These techniques differ in seven ways [19]:
1. Several cameras.
2. Response time and speed.
3. The physical features of the environment, such as the pace
of movement and illumination.
4. What accessories and clothing are required for use?
5. Low-level features like histograms, silhouettes, edges, mo-
ments, and regions are utilized.
6. Decide between two- and three-dimensional representa-
tions.
7. Is the representation of time accurate?
B. Glove-Based Approaches
Sayre Glove was designed by the Electronic Visualization Lab
in 1977, and it was the first data glove [20]. Researchers be-
lieve sign language influences gestures and can be used to cre-
ate computer instructions. Data glove approaches use sensors
to capture hand position, fingertips, acceleration, movement
velocity, orientation, and motion, enabling accurate computa-
tion of finger palm and hand configuration coordinates [21].
Sensors struggle with easy computer connection due to phys-
ical user connection, high cost, and unsuitability for virtual
reality environments, making them unsuitable for virtual real-