Remote control with accelerometer-based hand gesture recognition for interaction in digital TV

At the present, the digital TV allows the access to a greater amount of content and to execute interactive applications. The remote control used to control the digital TV systems is, in most cases, still solved by the traditional infrared remote control, which has become a limiting factor on the user interaction with the TV. This paper introduces the design and development of an interaction device for use in the context of digital TV in Argentina. The proposed device can be considered an evolution of the classic remote control, in which the functionality of hand gesture recognition is implemented as a natural and friendly interface for controlling digital TV systems of the home. A gestural dictionary of 20 types of gestures was adopted. The recognized gestures are translated into control commands for digital TV systems. As the hand gesture recognition is a pattern classification problem, two techniques based on artificial neural networks were explored, in order to compare results and to select the tool that best fits the problem in question. The pattern classifier design was described in detail, in order to properly select the hardware platform, fulfilling requirements of low-cost and fast execution of pattern classification algorithms. An interaction device of low-cost and excellent recognition precision was developed, for enhancing and enriching the user experience.