Back to results
US 9,807,416 B2Granted
Low-latency two-pass video coding
AssigneeGoogle Inc.
InventorAki Kuusela
Life of this patent
Estimating the term…
Abstract
Low-latency two-pass video coding may include identifying an input frame from an input video stream, determining a reduced frame from the input frame, the reduced frame having a size smaller than a size of the input frame, encoding the reduced frame using a first encoder, generating a plurality of encoding parameters based on encoding the reduced frame, generating an encoded frame by encoding the input frame using the first encoder and the plurality of encoding parameters, including the first encoded frame in an output bitstream, and storing or transmitting the output bitstream.
Claims17
A method comprising:
- identifying an input frame from an input video stream;
- determining a reduced frame from the input frame, the reduced frame having a size smaller than a size of the input frame;
- encoding the reduced frame using a first encoder, wherein encoding the reduced frame includes generating a plurality of encoding metrics;
- generating a plurality of encoding parameters based on the plurality of encoding metrics;
- generating an encoded frame by encoding the input frame using the first encoder and at least one encoding parameter from the plurality of encoding parameters;
- including the first encoded frame in an output bitstream;
- and
- storing or transmitting the output bitstream.
5 more elements
A method comprising:
- identifying an input frame from an input video stream, wherein the input frame includes an input block;
- determining a reduced frame from the input frame, the reduced frame having a size smaller than a size of the input frame, wherein the reduced frame includes a reduced block spatially corresponding to at least a portion of the input block;
- encoding, by a first encoder, the reduced frame, wherein encoding the reduced frame includes generating a plurality of encoding metrics, wherein generating the plurality of encoding metrics includes identifying a motion vector for encoding the reduced block;
- generating a plurality of encoding parameters based on the plurality of encoding metrics;
- generating, by the first encoder, an encoded frame by encoding the input frame based on at least one encoding parameter from the plurality of encoding parameters, wherein encoding the input frame includes, on a condition that the motion vector exceeds a motion threshold, using a motion search area for motion estimation for the input block, wherein the motion search area is identified based on the motion vector;
- including the encoded frame in an output bitstream;
- and
- storing or transmitting the output bitstream.
5 more elements
A method comprising:
- identifying an input frame from an input video stream;
- determining a reduced frame from the input frame, the reduced frame having a size smaller than a size of the input frame, wherein determining the reduced frame from the input frame includes cropping the input frame, scaling the input frame, or cropping and scaling the input frame;
- encoding, using a first encoder, the reduced frame, wherein encoding the reduced frame includes generating a plurality of encoding metrics;
- generating at least one encoding parameter based on the plurality of encoding metrics;
- generating, using the first encoder, an encoded frame by encoding the input frame using the at least one encoding parameter;
- including the first encoded frame in an output bitstream;
- and
- storing or transmitting the output bitstream.
5 more elements
Dependencies are inferred from the claim text.