Method and apparatus for detecting foreground windows
View Patent ↗A method, non-transitory computer readable medium and apparatus for detecting a window in a display are disclosed. For example, the method includes scanning pixels within each frame of a plurality of frames that is displayed via a sliding window, extracting one or more features from each pixel within the sliding window, applying a classification function to classify a subset of potential pixels comprising the each pixel within the sliding window as corner pixels based on the one or more features that match predefined features associated with the corner pixels and detecting the window based on the corner pixels and additional pixels within a boundary defined by the corner pixels.
1. A method for detecting a window in a display, comprising:
determining, by a processor, a color scheme of windows for an operating system using the window;
creating, by the processor, predefined features associated with a set of corner pixels for the color scheme that is deployed on the operating system, wherein the predefined features are represented as a triplet of a Haar-like feature comprising a type of feature, a color channel and a region defined by a left coordinate, a top coordinate, a right coordinate and a bottom coordinate;
scanning, by the processor, all pixels of each frame of a plurality of frames that is displayed via a sliding window that comprises a subset of pixels of the all pixels of the frame;
extracting, by the processor, the Haar-like feature of each pixel within the sliding window;
applying, by the processor, an Adaptive Boosting classification to select and combine a plurality of triplets associated with the Haar-like feature of a subset of potential pixels comprising the each pixel within the sliding window that have features that match the pre-defined features to generate a decision function that identifies corner pixels; and
detecting, by the processor, the window based on the corner pixels and additional pixels within a boundary defined by the corner pixels.
2. The method of claim 1 , further comprising:
detecting, by the processor, a plurality of windows based on the corner pixels; and
identifying, by the processor, a foreground window based on a relative position of each one of the plurality of windows.
3. The method of claim 2 , further comprising:
determining, by the processor, that the foreground window is not a desired foreground window; and
closing, by the processor, the foreground window.
4. The method of claim 2 , further comprising:
determining, by the processor, that the foreground window is not a desired foreground window; and
selecting, by the processor, a different window of the plurality of windows to be the foreground window.
5. The method of claim 1 , wherein the predefined features identified by the Haar-like feature comprise an edge.
6. The method of claim 5 , wherein the edge is defined by pixels having an average variance in a type of feature below a threshold.
7. A non-transitory computer-readable medium storing a plurality of instructions, which when executed by a processor, cause the processor to perform operations for detecting a window in a display, the operations comprising:
determining a color scheme of windows for an operating system using the window;
creating predefined features associated with a set of corner pixels for the color scheme that is deployed on the operating system, wherein the predefined features are represented as a triplet of a Haar-like feature comprising a type of feature, a color channel and a region defined by a left coordinate, a top coordinate, a right coordinate and a bottom coordinate;
scanning all pixels of each frame of a plurality of frames that is displayed via a sliding window that comprises a subset of pixels of the all pixels of the frame;
extracting the Haar-like feature of each pixel within the sliding window;
applying an Adaptive Boosting classification to select and combine a plurality of triplets associated with the Haar-like feature of a subset of potential pixels comprising the each pixel within the sliding window that have features that match the pre-defined features to generate a decision function that identifies corner pixels; and
detecting the window based on the corner pixels and additional pixels within a boundary defined by the corner pixels.
8. The non-transitory computer-readable medium of claim 7 , the operations further comprising:
detecting a plurality of windows based on the corner pixels; and
identifying a foreground window based on a relative position of each one of the plurality of windows.
9. The non-transitory computer-readable medium of claim 8 , the operations further comprising:
determining that the foreground window is not a desired foreground window; and
closing the foreground window.
10. The non-transitory computer-readable medium of claim 8 , the operations further comprising:
determining that the foreground window is not a desired foreground window; and
selecting a different window of the plurality of windows to be the foreground window.
11. The non-transitory computer-readable medium of claim 7 , wherein the pre-defined features identified by the Haar-like feature comprise an edge.
12. The non-transitory computer-readable medium of claim 11 , wherein the edge is defined by pixels having an average variance in a type of feature below a threshold.