Design Of The Gesture Interaction Systems For Smart TV

The hand control technology is a harmonious interaction technic widely used by large TV manufacturers on the smart TV. However, as gestures are complicated and changing quickly while the home provides complex lighting and background conditions, the effect of gesture tracking of the manufactures is not satisfying while the levels of their instruction set are uneven. Therefore, it is highly significant to design a stable and pragmatic hand control system to accelerate the development of this industry.The author studied and designed a new hand control interaction system for smart TV based on 2D colour camera. The whole hand control interaction system consists of three phases: gesture detection, gesture tracking and gesture recognition. On the basis of the current technology, this paper has made improvements on the gesture tracking algorithm, gesture recognition algorithm, and instruction set design.(1) On gesture tracking. In order to adapt to the non-rigid deformation and the complex lighting and background at home, based on the incremental subspace tracking model, the tracking algorithm introduces online feature selection into the tracking module and add weight module into subspace to enhance the stability and robustness of the gesture tracking. And the test of this paper proves its accuracy and robustness are better than the current algorithm.(2) On the gesture recognition. The traditional gesture instruction set designs are unnatural with only one algorithm and cannot be developed according to different gestures. Sticking to the principle of accurate, fast, easy to operate and understand, this paper designs a new instruction set with 11 gestures, including a new gesture "fist" recognition algorithm designed based on light streamer, which improves the "track match" recognition algorithm. Apart from this, the author also developed a static instruction based on Adaboost algorithm. Moreover, this paper has also carried out tests to prove the recognition rate and accuracy of the new algorithm are better and more pragmatic in normal conditions.(3) On the design of the system. For the smart TV end, in order to effectively distribute the process resources to balance the complexity and the speed of the algorithm, this paper adopts high-profile PC processor while the instruction response is processed by the smart TV system connected with it via network. For problem of tracking loss, this paper designed a rechecking strategy, which is quite useful in real application.