×

Manipulation input device which detects human hand manipulations from captured motion images

  • US 8,378,970 B2
  • Filed: 10/20/2008
  • Issued: 02/19/2013
  • Est. Priority Date: 10/22/2007
  • Status: Active Grant
First Claim
Patent Images

1. A manipulation input device, comprising:

  • a plurality of different kinds of picture-taking apparatuses which include a visible light camera;

    a plurality of hand region detectors that detect a region of a human hand from a motion image taken with each of the plurality of picture-taking apparatus;

    a plurality of hand manipulation determination units that determine a hand manipulation from a shape and a motion of the detected hand region;

    a hand manipulation determination and selection unit that selects a single result from results determined by the plurality of hand manipulation determination units; and

    a selection menu representation unit that notifies a user of a menu selected based on the manipulation selected by the hand manipulation determination and selection unit,wherein at least one of the hand region determination units comprises;

    a skin tone region extraction unit that extracts a region of a skin tone from a motion image taken with the visible light camera;

    a difference region extraction unit that extracts region data differing between current motion image data and background image data in an image taken with the visible light camera;

    a binary correction unit that digitizes image data of the region extracted by the skin tone region extraction unit and the difference region extraction unit, to then correct the digitized image data;

    a distance calculator that calculates minimum distances from inner points of the binary image region obtained by the binary correction unit to an edge thereof, to detect as the center of a hand a distance maximum point where the calculated values become a maximum one;

    a center weighted correction unit that adds to the distance value of the distance maximum point a correction value such that the closer the position of the distance maximum point is to the center of an imaging screen, the larger the correction value, and the closer the position thereof is to the end portion of the imaging screen, the smaller the correction value;

    a candidate hand region detector that detects a hand region from a binary image region that includes a point where the distance value of the distance maximum point to which the correction value is added is a maximum;

    an edge length and area calculator that calculates an edge length, and an area, of the candidate hand region; and

    a hand region determination unit that determines whether or not the candidate hand region is a hand from the calculated edge length and area.

View all claims
  • 1 Assignment
Timeline View
Assignment View
    ×
    ×