Automatic Hand-Pose Trajectory Tracking System Using Video Sequences
Yuan‐Hsiang Chang, Chen‐Ming Chang · InTech eBooks · 2010
Introduction 1.1 BackgroundHand-pose is one of the most important communication tools in human's daily life.In a situation when people from different countries are trying to communicate, they may be able to roughly express their thought through hand-poses or hand-gestures.During an oral presentation, a presenter may use hand-poses as an auxiliary tool to convey his or her idea for technical communication.In practice, the use of "sign language" is a typical example that provides an effective mechanism for exchanging information among deaf people and the hearing society.In essence, "sign language" can be considered as a combination of many hand-poses that define the actual information.With the continuous advances of speech, image, and video processing techniques, humanmachine interaction is constantly making progress over the past decades.For example, speech recognition systems (Cooke et al., 2001;Gales, 1998) are designed to recognize and translate human's speech to text.Handwritten recognition systems (Liu et al., 2003;Palacios & Gupta, 2002;Zhai & Kristensson, 2003) are designed to recognize and translate human's handwritten to text.A palmprint identification system (Zhang et al. 2003) is a biometric approach to recognize human's palmprint for personal identification.A safety vehicle system (Trivedi et al., 2007) can be used to estimate the situation around a vehicle and convey information to warn the driver for potential dangers such that the vehicle safety could be improved.In summary, many of these systems have been found to be feasible in enhancing the human-machine interaction and integrated in daily applications (e.g., cell phones, notebooks, security system, etc.). Related researchDuring the past years, researchers have proposed novel methods for the classification or recognition of hand-poses or hand-gestures.The techniques can be divided into two main categories: image-based approaches and glove-based approaches.The image-based approaches are generally designed to use images as inputs to the system for the hand-pose recognition.In contrary, the glove-based approaches are designed with a special hardware installment and/or sensors (e.g., a data glove) as inputs to the system for the hand-pose recognition.The two approaches can be described as follows.