Feature-based video compression
First Claim
1. A computer implemented method of processing video data formed of a series of video frames comprising:
- using a first video encoding process an d feature-based encoding process, processing the video frames by;
encoding the video frames with the first video encoding process;
processing one or more of the frames to detect one or more instances of a feature by searching the one or more frames for a region of pels having coherency and computational complexity as compared to other pels in the one or more frames;
modeling variation of the feature instance relative to other instances of the feature to create a feature-based encoding of the feature instance, wherein modeling variation of the feature instance includes;
identifying other instances of the feature in other frames of the video series;
aggregating two or more of the instances of the feature into a set;
processing the feature instances in the set to create the models for the feature based encoding;
comparing compression efficiency of the feature-based encoding of the feature instance relative to an encoding of the feature instance resulting from the first video encoding process;
determining from the comparison which encoding enables greater compression efficiency; and
using the results of the comparing and determining step, applying feature-based encoding to portions of one or more of the video frames, and applying conventional encoding to other portions of the one or more video frames.
1 Assignment
0 Petitions
Accused Products
Abstract
Systems and methods of processing video data are provided. Video data having a series of video frames is received and processed. One or more instances of a candidate feature are detected in the video frames. The previously decoded video frames are processed to identify potential matches of the candidate feature. When a substantial amount of portions of previously decoded video frames include instances of the candidate feature, the instances of the candidate feature are aggregated into a set. The candidate feature set is used to create a feature-based model. The feature-based model includes a model of deformation variation and a model of appearance variation of instances of the candidate feature. The feature-based model compression efficiency is compared with the conventional video compression efficiency.
-
Citations
46 Claims
-
1. A computer implemented method of processing video data formed of a series of video frames comprising:
-
using a first video encoding process an d feature-based encoding process, processing the video frames by; encoding the video frames with the first video encoding process; processing one or more of the frames to detect one or more instances of a feature by searching the one or more frames for a region of pels having coherency and computational complexity as compared to other pels in the one or more frames; modeling variation of the feature instance relative to other instances of the feature to create a feature-based encoding of the feature instance, wherein modeling variation of the feature instance includes; identifying other instances of the feature in other frames of the video series; aggregating two or more of the instances of the feature into a set; processing the feature instances in the set to create the models for the feature based encoding; comparing compression efficiency of the feature-based encoding of the feature instance relative to an encoding of the feature instance resulting from the first video encoding process; determining from the comparison which encoding enables greater compression efficiency; and using the results of the comparing and determining step, applying feature-based encoding to portions of one or more of the video frames, and applying conventional encoding to other portions of the one or more video frames. - View Dependent Claims (2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34)
-
-
35. A computer system for processing video comprising:
-
a hybrid encoder configured to encode at least a portion of one or more video frames using a first video encoding process based on motion compensated prediction and encoding one or more other portions of the video frames using another video encoding process based on feature-based compression, the encoder configured to; identify in a plurality of video frames instances of a feature, where the feature is a spatially continuous region of pels that are in close spatial proximity to each other having coherency and exhibiting complexity; determine that an encoding of the feature instances using models from a feature-based encoding process provides greater compression efficiency relative to an encoding of the features instances resulting from the first video encoding process; using the results of the and determining step, applying feature-based encoding to portions of one or more of the video frames, and applying conventional encoding to other portions of the one or more video frames; wherein applying the feature based encoding to one or more of the feature instances by modeling their variation in the plurality of the video frames, wherein modeling variation of the feature instance includes; identifying other instances of the feature in other frames of the video series; aggregating two or more of the instances of the feature into a set; processing the feature instances in the set to create the models for the feature based encoding. - View Dependent Claims (36, 37, 38, 39, 40, 41, 42)
-
-
43. A non-transitory computer readable medium storing instructions so as when executed to cause one or more computer processors to encode video frames by:
-
processing the video frames with a first video encoding process that uses motion compensated prediction; searching or more of the frames for an instance of a feature that includes a spatially continuous region of pels in close spatial proximity to each other having coherency and exhibiting complexity; identifying other instances of the feature in other frames of the video series; aggregating two or more of the instances of the feature into a set; processing the feature instances in the set to create the models for the feature based encoding; comparing compression efficiency of a feature based encoding of the feature instance relative to an encoding of the feature instance resulting from the first encoding process; and determining from the comparison which encoding enables greater compression efficiency; and using the results of the comparing and determining step, applying feature-based encoding to portions of one or more of the video frames, and applying conventional encoding to other portions of the one or more video frames.
-
-
44. A computer implemented method for processing video data comprising:
-
processing portions of encoded video frames including one or more encoded features, the one or more encoded features including portions of features encoded using a feature-based encoding process and portions of features encoded using a motion compensated encoding process; separating encoded features from non-feature portions of the video frames; separating the portions of features encoded using feature-based encoding from portions of features encoded using a motion compensated encoding process; decoding the non-feature portions of the video and the portions of features encoded using a motion compensated encoding process using a motion compensated decoding process; decoding the portions of features encoded using feature-based encoding using encoded feature parameters and feature models generated by the decoder; and compositing the decoded non-feature portions, the decoded motion compensated encoded feature portions and the decoded feature-based encoded feature portions to generate a decoded video signal; wherein generating feature models by the decoder includes; filtering on or more of the decoded frames to detect an instance of a feature that is a spatially continuous region of pels that are in close spatial proximity to each other, the regions of pels having coherency and exhibiting complexity; and identifying other instances of the feature in other decoded frames of the video series; aggregating two or more of the instances of the feature into a set; processing the feature instances in the set to create the models for the feature based decoding.
-
-
45. A computer system for processing video comprising:
-
a hybrid decoder configured to decode portions of the video frames including one or more encoded features using a first video encoding process based on motion compensated prediction and to decode other portions of the video frames using another video decoding process based on feature based compression by; separating encoded features from non-feature portions of the video frames; separating the portions of features encoded using feature-based encoding from portions of features encoded using the first encoding process; decoding the non-feature portions of the video and the portions of features encoded using the first encoding process using a motion compensated decoding process; decoding the portions of features encoded using feature-based encoding using encoded feature parameters and feature models generated by the decoder; and compositing the decoded non-feature portions, the decoded first encoding process encoded feature portions and the decoded feature-based encoded feature portions to generate a decoded video signal; wherein generating feature models by the decoder includes; identifying in a plurality of decoded video frames instances of a feature where the feature is a spatially continuous region of pels that are in close spatial proximity to each other having coherency and exhibiting complexity; and identifying other instances of the feature in other decoded frames of the video series; aggregating two or more of the instances of the feature into a set; processing the feature instances in the set to create the models for the feature based decoding.
-
-
46. A non-transitory computer readable medium storing instructions so as when executed to cause one or more computer processors to decode video frames by:
-
processing the encoded video frames including one or more encoded features, with a first video decoding process based on motion compensated prediction and to decode other portions of the video frames using another video decoding process based on feature based compression by; separating encoded features from non-feature portions of the video frames; separating the portions of features encoded using feature-based encoding from portions of features encoded using a motion compensated encoding process; decoding the non-feature portions of the video and the portions of features encoded using a motion compensated encoding process using a motion compensated decoding process; decoding the portions of features encoded using feature-based encoding using encoded feature parameters and feature models generated by the decoder; and compositing the decoded non-feature portions, the decoded motion compensated encoded feature portions and the decoded feature-based encoded feature portions to generate a decoded video signal; wherein generating feature models by the decoder includes; identifying, in one or more of the decoded frames, an instance of a feature that is a spatially continuous region of pels in close spatial proximity to each other having coherency and exhibiting complexity; and identifying other instances of the feature in other decoded frames of the video series; aggregating two or more of the instances of the feature into a set; processing the feature instances in the set to create the models for the feature based decoding.
-
Specification