02-05-2011, 03:31 PM
[attachment=13253]
Abstract—In this paper we present an automatic hand gesture
recognition system operating on video stream. The system
consists of two modules: hand gesture detection module and
hand gesture recognition module. The detection module could
accurately locate the hand regions with a blue rectangle; this is
mainly based on Viola-Jones method, which is currently
considered the fastest and most accurate learning-based
method for object detection. In the recognition module, the Hu
invariant moments feature vectors of the detected hand gesture
are extracted and a Support Vector Machines (SVMs)
classifier is trained for final recognition, due to its high
generalization performance without the need to add a priori
knowledge. The performance of the proposed system is tested
through a series of experiments and a simple Human-
Computer Interaction application based on hand gesture
recognition method is finally developed.
Keywords- hand gesture detection; hand gesture recognition;
Viola-Jones method; Hu invariont moments; SVMs; Human-
Computer Interaction
I. INTRODUCTION
Vision-based hand gesture recognition has drawn
considerable attention from researchers in recent years. An
automatic hand gesture recognition system will find many
applications in Human-Computer Interaction area. Hand
gesture detection is a fundamental step in the practical
application process of this system, which requires the ability
to accurately segment the hand from the background. Due to
the difficulty of this task, early systems usually require
markers or colored gloves to make the detection easier [1]
[2]. However, these methods often bring much
inconvenience for the Human-Computer Interaction process.
Moreover, the current research is mainly focused on
detecting the bare hand and recognizing hand gestures
without any markers and gloves. To solve this problem, the
proposed system adopts Viola-Jones method [3] for hand
gesture detection, which is currently considered the fastest
and most accurate learning-based object detection method.
Based on Viola-Jones method, the proposed detection
module could accurately find the candidate hand gesture
region and locate it with a blue rectangle.
Vision-based recognition of hand gesture is also a
challenging problem in Human-Computer Interaction areas.
One common way to classify gesture recognition techniques
is by approach, such as analyzing feature, or determining an
underlying model. This approach typically involves predetermining
a set of common discriminating features,
estimating covariance during a training process, and using a
linear discriminator [4] to classify the gestures. As a
reference, this recognition approach is applied to develop the
recognition module of the proposed system. Based on the
detection module, the Hu invariant moment features of the
detected hand gesture are extracted and a Support Vector
Machines (SVMs) classifier is trained for classification.
Combining relevant feature extraction and Support Vector
Machines (SVMs) can significantly increase the robustness
of hand gesture recognition in real-time. The main
framework of the proposed system is shown
