摘要

Considering the low accuracy instability factors in case of the hand shape change and background change, an approach to track dynamic deformable hand gesture based on joint spatial-feature space is proposed. Using the hand gesture pixel feature information of the spatial domain and color space, the kernel probability density of hand gesture's sample set is calculated, and the moving hand gesture's position in spatial is located by Mean-Shift. The experiments show that this approach can track dynamic hand gesture accurately in real time, and the time complexity is low, so it guarantees the stability and continuity of the hand tracking.

  • 出版日期2012

全文