×

Skin tone and feature detection for video conferencing compression

  • US 8,588,309 B2
  • Filed: 04/07/2010
  • Issued: 11/19/2013
  • Est. Priority Date: 04/07/2010
  • Status: Active Grant
First Claim
Patent Images

1. A processor programmed to perform a video compression method, the method comprising:

  • determining with the processor any first pixels in a frame having color in a predetermined tone region;

    determining with the processor any second pixels in the frame that are part of at least one facial feature;

    for each pixel in the frame;

    assigning a first value to the respective pixel if a location of the pixel is within a specified threshold distance to any of the first pixels, but the pixel is not one of the second pixels;

    assigning a second value to the respective pixel if the pixel is one of the second pixels but is not located within a specified threshold distance to any of the first pixels;

    assigning a third value to the respective pixel if the location of the pixel is within a specified threshold distance to any first pixel and the pixel is a second pixel;

    assigning a fourth value to the respective pixel if the pixel is not located within a specified threshold distance to any first pixel and not one of the second pixels;

    averaging the assigned pixel values on a block by block basis within the frame to generate respective block scores;

    compressing with the processor the video frame using coding parameters selected based on the block scores.

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