feature processing and modelling for 6D motion gesture database.....
1. FEATURE PROCESSING &
MODELLING FOR 6D MOTION
GESTURE RECOGNITION
1
GUIDED BY PRESENTED BY
Mrs. Seena. M .K Jeeva John
Asst. Prof., ECE S7 ECE (035)
2. ABSTRACT
• A 6D motion gesture is represented by a 3D spatial trajectory and
augmented by another three dimensions of orientation. Using
different tracking technologies, the motion can be tracked explicitly
with the position and orientation or implicitly with the acceleration
and angular speed. This deals with the relative effectiveness of
various feature dimensions for motion gesture recognition in both
user-dependent and user-independent cases. A statistical feature-
based classifier and an HMM-based recognizer, which offers more
flexibility in feature selection and achieves better performance in
recognition accuracy than another systems. Motion gesture database
contains both explicit and implicit motion information which allows
comparing the recognition performance of different tracking signals
on a common ground. This gives an insight into the attainable
recognition rate with different tracking devices, which is valuable
for the system designer to choose the proper tracking technology.
3. INTRODUCTION
• “Dimension” is a property of space(1D,2D, 3D..)
• A straight line has 1dimension( 1 coordinate)
• A parallelogram has 2 dimension(2 coordinate)
• A parallelepiped has 3dimension.(3 coordinates)
• Similar way, a 6D in any space that has 6 coordinates.
• The implementation of a gesture-control interface contains two key
components:
GESTURE RECOGNITION &
MOTION TRACKING
• 6D motion gestures are used for recognizing the gesture.
• The 6D motion gesture is represented by,
3D SPATIAL TRAJECTORY &
OTHER 3 DIMENSIONS OF ORIENTATION
4. What is gesture?
• GESTURE is a meaningful body movement expressed by a subject.
• Purpose of human gestures:
conversational, controlling, commanding, manipulative, and
communicative.
Two types;
Natural gestures
Sign language gestures
Natural gestures
• Free form
• Can occur in any order, dimension, shape etc…
• The gesture performed by different individuals can vary dynamically.
• Sign language gestures
• It has a defined grammar.
• Dynamic motions, i.e., motion gestures are commonly used for
‘command & control’ process.
5. Why gestures?
• It is quite Natural.
• Most Efficient.
• High accuracy.
• Gestures can be culture specific.
6. Gesture recognition
• Two extreme cases are,
o User dependent gesture recognition.
o User independent gesture recognition.
• Major approaches for analyzing or recognizing a gesture includes,
• Dynamic Time Warping (DTW)
• Neural Networks (NNs)
• Hidden Markov Models (HMMs)
• Data-driven template matching.
• Statistical feature based classifiers.
• Statistical feature based classifiers are also called as rubine classifier
7. DTW(Dynamic Time Warping…)
Dynamic time warping (DTW) is an algorithm for measuring similarity
between two sequences which may vary in time or speed.
DTW can be useful for personalized gesture recognition.
These recognizers are simple to implement, computationally
inexpensive, and require only a few training samples to function properly.
A significant amount of templates are needed to cover the range of
variations.
When a large set of training samples is hard to collect, we use the DTW
technique.
8. A child being sensed by a simple gesture
recognition algorithm detecting hand location
and movement
9. HMM BASED CLASSIFIER
• Most commonly used classifier for 6D gesture recognition.
• The HMM is efficient at modelling a time series with and
temporal variations, and has been successfully applied to
gesture recognition.
• The features (observations)for the HMMs vary, including the
position, moving direction, acceleration, etc.
• Normalization procedure is used specifically for the explicit
and implicit motion data.
10. 6DMG: 6D MOTION GESTURE
DATABASE
• Here we define a total of 20 gestures.
• Including swiping motions in eight directions, poke gestures that swipe
rapidly forth and back in four directions, v-shape, x shape, clockwise and
counter clockwise circles in both vertical and horizontal planes, and wrist
twisting (roll).
• Wii Remote Plus (Wiimote) for the inertial measurement of the
acceleration and angular speed.
• There are no mirror gestures, which means the direction and rotation are
the same for both right- and left-handed users.
• Wiimote used here, which works as a position tracker.
• The tracking device provides both explicit and implicit 6D spatio-temporal
information sampled at 60 Hz, including the
position, orientation, acceleration, and angular speed.
13. What is wiimote
• It is a REMOTE.
• The body of the Wii Remote measures 148 mm (5.8 in)
long, 36.2 mm (1.43 in) wide, and 30.8 mm (1.21 in) thick.
• Feature of the Wii Remote is its motion sensing capability.
• Wiimote allows the user to interact with and manipulate items
on screen via gesture recognition and pointing through the use
of accelerometer and optical sensor technology.
15. Block diagram for gesture recognition
• These are all the techniques that are used for recognizing a
gesture.
RECOGNIZED
GESTURE
GESTURE RECOGNITION
HMMDTW
6D
MG
16. Motion tracking
• Motion tracking is used to capture the motion before performing gesture
recognition.
Tracking an object in space actually requires six dimensions:
Three for translation
Three for rotation
• Methods used for tracking motion…
Two methods used for tracking the motion ,
Explicit method.
Implicit method.
• The motion can be tracked explicitly with the position and orientation.
• Implicitly with the acceleration and angular speed.
17. Technologies for motion tracking
• There are several technologies for motion tracking.
Vision based technique.
&
Tracker based technique.
The tracker based technique ( used for sensing) can be
divided into,
Optical sensing
&
Inertial sensing
18. VISION BASED TECHNIQUE
• This provides more natural and unencumbered (free form) interaction.
MONOCULAR IMAGES OR VIDEOS:
Can extract the projected 2D trajectory and the orientation
DEPTH CAMERA:
estimates the depth in a rougher scale.
STEREO OR MULTI VIEW CAMERAS:
To track a full 3D motion
XBOX 360 KINECT:
Used for tracking a human body in 3D
20. TRACKER BASED TECHNIQUE
• Tracker-based techniques achieve more precise motion
tracking at the expense of requiring the user to wear certain
equipment.
• A motion tracking system most often derives estimate of
motion information from magnetic, acoustic, inertial, or
optical sensors.
• From this optical sensors and inertial sensors are most
commonly used.
21. Optical sensors
• An optical sensor is a device that converts light rays into electronic signals.
• It is Similar to a photo resistor
• Tracks the global orientation and position.
• i.e. Optical sensors track either active or reflective markers and provide
accurate motion tracking results at a relatively high speed.
• Here at least two pairs of the tracker-sensor relationship are needed for
valid triangulation to determine the position of one tracker.
• Advantages
1. High sensitivity.
2. Small size and longer lifetime.
3. High intensity.
23. Inertial sensor
• Inertial sensors are commonly referred to the MEMS
(microelectronic mechanical systems)
• It includes, accelerometers and gyroscopes in chip form.
• The accelerometer measures the accelerations in the device-
wise coordinates.
• Gyroscope measures the angular speed and the orientation.
• Both has the ability to sense the vibration, rotation, tilt etc…
27. CONCLUSION
• Gestures can be a natural and intuitive way for interaction, and we are
especially interested in motion gestures rendered by the hand or handheld
device in free space without regard to the posture, finger or body
movements.
• With different tracking technologies, the affordable motion information
varies, which can be the position, orientation, acceleration, and angular
speed. Although motion gestures are usually defined by the spatial
trajectory, other kinematic properties still contain information to distinguish
the gestures.
• Two techniques have been used, motion tracking and gesture recognition.
28. REFERENCES
• G. Welch and E. Foxlin, “Motion tracking: no silver bullet, but a
respectable arsenal,” Computer Graphics and Application
• R. Teather, A. Pavlovych, W. Stuerzlinger, and I. MacKenzie, “Effects of
tracking technology, latency, and spatial jitter on object
movement, "Proceedings of IEEE Symposium on 3D User Interfaces
• M. Chen, G. AlRegib, and B.-H. Juang, “6dmg: A new 6d motiongesture
database,” in Proceedings of the third annual ACM conferenceon
Multimedia systems
• J. Ruiz, Y. Li, and E. Lank, “User-defined motion gestures for mobile
interaction,” in Proceedings of the 29th international conference on Human
factors in computing systems