Method of extracting logo mask from video stream having logo

A method of extracting a logo mask from a video stream having a logo, the method comprising:

• receiving the video stream including a plurality of frames,
• generating sequentially a plurality of online-weighted-average frames using the video stream;
• and
• generating a logo-boundary mask using a current frame among the plurality of frames and a previous online-weighted-average frame among the plurality of online-weighted-average frames, wherein the logo-boundary mask includes a logo-boundary pixel and a non-logo-boundary pixel;
• and
• filling the logo-boundary mask to generate the logo mask according to a plurality of logo-boundary scores, wherein a first logo-boundary score of the non-logo-boundary pixel is determined by comparing an intensity of the non-logo-boundary pixel in the current frame and an average intensity of the non-logo-boundary pixel in a current online-weighted-average frame among the plurality of online-weighted-average frames,
• wherein the logo mask includes a plurality of entries, each of which represents whether a pixel of the video stream is a logo pixel or a non-logo pixel.

A method of detecting a logo from a video stream, the method comprising:

• receiving the video stream including a plurality of frames;
• generating a current edge map using a current frame among the plurality of frames;
• generating a current online-weighted-average frames by recursively adding an intensity of the current frame of the video stream to an intensity of a previous online-weighted-average frame;
• generating an average edge map using the current online-weighted-average frame;
• and
• generating a logo-boundary-score map using the current edge map and the average edge map.