Object extraction
AssigneeArcSoft, Inc.
InventorsBin Zhao, Na Li
Life of this patent
Estimating the term…
Abstract
A video sequence is processed to extract one or more objects. The extracted objects are used to create a new video sequence. In particular, a user is able to select a background and/or an audio recording. Alternatively, the object extraction system automatically selects a background and/or audio recording. The extracted objects are integrated into the selected background and/or the audio recording is added to form a new video sequence.
Claims66
A method for extracting a foreground object from a background, comprising:
- (a) when a value component of a pixel in a non-static frame exceeds a mean value of the value component for the pixel across multiple frames by a first value, classifying the pixel as foreground;
- (b) when the pixel is in shadow, classifying the pixel as background;
- (c) when the pixel is under different illumination and each usable component of the pixel exceeds a mean value of that component by a second value, classifying the pixel as background;
- and
- (d) when the pixel is not classified by acts (a), (b), or (c), classifying the pixel as foreground.
A method for processing a sequence of frames, comprising:
- extracting one or more objects from a frame by classifying each pixel of the frame as being background or foreground based on a statistical model of a background in the frame, a type model of each pixel, shadow conditions, and illumination conditions;
- selecting a new background;
- selecting an audio recording;
- and
- integrating the extracted objects with the new background and audio recording.
A system, comprising:
- a computer including a memory;
- a sequence of frames stored in the memory;
- and
- a computer program executable by the computer, wherein the computer program comprises instructions for:
- (a) when a value component of a pixel in a non-static frame in the sequence of frames exceeds a mean value of the value component for the pixel across multiple frames in the sequence of frames by a first value, classifying the pixel as foreground;
- (b) when the pixel is in shadow, classifying the pixel as background;
- (c) when the pixel is under different illumination and each usable component of the pixel exceeds a mean value of that component by a second value, classifying the pixel as background;
- and
- (d) when the pixel is not classified by acts (a), (b), or (c), classifying the pixel as foreground.
A system, comprising:
- means for extracting one or more objects from a frame by classifying each pixel of the frame as being background or foreground based on a statistical model of a background in the frame, a type model of each pixel, shadow conditions, and illumination conditions;
- means for selecting a new background;
- means for selecting an audio recording;
- and
- means for integrating the extracted objects with the new background and audio recording.
A computer readable storage medium encoded with a computer program executable by a computer, wherein the computer program comprises instructions for:
- (a) when a value component of a pixel in a non-static frame exceeds a mean value of the value component for the pixel across multiple frames by a first value, classifying the pixel as foreground;
- (b) when the pixel is in shadow, classifying the pixel as background;
- (c) when the pixel is under different illumination and each usable component of the pixel exceeds a mean value of that component by a second value, classifying the pixel as background;
- and
- (d) when the pixel is not classified by acts (a), (b), or (c), classifying the pixel as foreground.
A computer readable storage medium encoded with software instructions, wherein the computer program further comprises instructions for:
- extracting one or more objects from a frame by classifying each pixel of the frame as being background or foreground based on a statistical model of a background in the frame, a type model of each pixel, shadow conditions, and illumination conditions;
- selecting a new background;
- selecting an audio recording;
- and
- integrating the extracted objects with the new background and audio recording.
Dependencies are inferred from the claim text.