US 6,553,150 B1 · Image sequence compression featuring independently coded regions

Abstract

In a MPEG or other video coding system, regions of each frame may be independently coded for later, independent extraction or processing directly from a compressed bit stream. An encoder/transcoder receives raw video, standard compressed video or compressed video already having independently coded regions (“ICRs”) that are to be edited in some manner. The encoder/transcoder permits user creation of regions, and provides automatic tracking features to identify and select those objects or regions through multiple frames (notwithstanding object movement). The encoder/transcoder re-uses as much compressed input data as is available in generating an output, and so, may be used for real-time encoding and editing processes. To this effect, the encoder/transcoder re-uses original bit stream data as well as original or new motion vector data in compiling an output, or any mix of them as appropriate.

Claims21

Dependent claims

In a video compression system, an improvement comprising:

identifying at least two data sets in a video frame that are to be independently encoded;
dividing the video frame into image blocks, and associating each image block with a data set;
motion-compensating the image blocks to generate motion-compensated blocks, but constraining motion vectors for each image block to point only to reference data within the same data set as the image block;
encoding the motion-compensated blocks in a sequential order and beginning a new image slice each time a next motion-compensated block in the sequential order is associated with a different data set;
and
producing an output bitstream where image slices are ordered in correspondence with the sequential order of the image blocks.

3 more elements

A method of compressing video to have independently-encoded regions that may be separately extracted from an output video signal without completely decoding the output video signal, said method comprising:

identifying at least two data sets in a video frame that are to be independently encoded;
dividing the video frame into a sequence of image blocks that are to be motion-compensated, and associating each image block exclusively with one data set;
performing motion-compensation upon the image blocks in a constrained manner that limits motion vector dependence to reference data within the same data set;
encoding motion-compensated blocks into substantially non-overlapping image slices, including starting a new image slice each time a next motion-compensated block represents a different data set than the immediately previous motion-compensated block in the sequence;
and
producing the output signal from the image slices.

3 more elements

An apparatus, comprising:

a region selection device that identifies subsets of a video sequence that are to be independently encoded from one another;
a motion compensation device that encodes blocks of video data for a frame into motion vector and residual format, but that limits motion compensation to use of reference data associated with the same subset;
an encoding device that builds an output bitstream having discrete image slices separately parsable from the output bitstream, the encoding device sequentially encoding motion-compensated blocks in a manner that each image slice represents motion-compensated blocks associated with a single subset, and that a new image slice is defined each time a next-motion compensated block in the sequence represents a change in subsets that are to be independently encoded.

Dependencies are inferred from the claim text.

Image sequence compression featuring independently coded regions

Life of this patent

Image sequence compression featuring independently coded regions

Life of this patent

Abstract

Claims21