Hybrid video resolution method and apparatus
View Patent ↗In some respects, the disclosure is directed to methods and systems for processing signals. A device can include a first circuit configured to process video information for a resolution or scaling operation and provide first data, a second circuit configured to process the video information for the resolution or scaling operation and provide second data, and a third circuit configured to receive the first data and the second data and provide third data according to the resolution or scaling operation, the first circuit comprising a first machine learning engine.
1 . A device, comprising:
a first circuit configured to process video information for a resolution or scaling operation and provide first data, the first circuit comprising a first machine learning engine;
a second circuit configured to process the video information for the resolution or scaling operation and provide second data;
a calculator configured to provide a value of probability that at least a portion of the video information comprises an invariant; and
a third circuit configured to receive the first data, the second data, and the value of probability, and blend the first data and the second data according to the value of probability.
2 . The device of claim 1 , wherein the third circuit comprises a blender configured to blend the first data and the second data in response to the value of probability.
3 . The device of claim 1 , wherein the calculator is configured to receive the video information.
4 . The device of claim 3 , wherein the calculator comprises a second machine learning engine.
5 . The device of claim 1 , wherein the calculator is configured to determine the value of probability using a distance to a plane.
6 . The device of claim 1 , wherein the first circuit is configured to process the video information for a super resolution operation.
7 . A system for converting pixel data from a first resolution to a second resolution or from a first scale to a second scale, the system comprising:
a first path comprising a machine learning engine;
a second path comprising a calculator, the calculator configured to provide a value related to a likelihood of presence of an invariant using a distance to a plane;
a third path comprising a converter;
an input configured to provide input pixel data to the first path, the second path, and the third path; and
a blender configured to combine first output pixel data from the first path and second output pixel data from the third path in response to the value.
8 . The system of claim 7 , wherein the machine learning engine is configured to convert the input pixel data at a first, lower resolution to the first output pixel data at a second higher resolution.
9 . The system of claim 7 , wherein the converter comprises a non-machine learning converter.
10 . The system of claim 7 , wherein the converter comprises a linear converter.
11 . The system of claim 7 , wherein the calculator is configured to use a machine learning technique to provide the value.
12 . A device, comprising:
a first circuit configured to process input data for a resolution or scaling operation and provide first output data, the first circuit comprising a first machine learning engine;
a second circuit configured to process the input data for the resolution or scaling operation and provide second output data;
a calculator configured to provide a value of probability that at least a portion of the input data comprises invariant data; and
a third circuit configured to receive the first output data, the second output data, and the value of probability, and blend the first output data and the second output data according to the value of probability.
13 . The device of claim 12 , wherein the third circuit comprises a blender configured to blend the first output data and the second output data in response to the value of probability.
14 . The device of claim 12 , wherein the invariant data corresponds to a group of pixels in the input data having a constant luminance value.
15 . The device of claim 12 , wherein the calculator comprises a second machine learning engine configured to provide the value of probability.
16 . The device of claim 12 , wherein the calculator is configured to provide the probability using a distance to a plane in a group of pixels in the input data.
17 . The device of claim 12 , wherein the first circuit is configured to process the input data for a super resolution operation.
18 . The device of claim 12 , wherein the input data is in a vector space format.