Global motion constrained motion vector in inter prediction
The decoder employs a simplified global motion model to address inefficiencies in inter prediction, enhancing encoding efficiency and decoding performance by reducing computational complexity.
Patent Information
- Authority / Receiving Office
- HK Β· HK
- Patent Type
- Applications
- Current Assignee / Owner
- DOLBY INTERNATIONAL AB
- Filing Date
- 2026-05-22
- Publication Date
- 2026-07-10
AI Technical Summary
Existing video encoding and decoding systems face inefficiencies in inter prediction due to the complexity of motion vectors, particularly when global motion is involved, leading to increased computational resources and reduced encoding efficiency.
A decoder is configured to utilize a global motion model with reduced complexity for each block, enabling efficient decoding by extracting a header from a bitstream that characterizes the global motion parameters.
Enhances encoding efficiency by reducing computational complexity and improving decoding performance through the use of a simplified global motion model.
Smart Images
Figure 00000000_0000_ABST
Abstract
Description
Abstract A decoder includes circuitry configured to receive a bitstream, extract a header associated with a current frame and including a signal characterizing that global motion is enabled and further characterizing parameters of a global motion model, and decoding the current frame, the decoding including using a motion model for each current block having a complexity that is less than or equal to a complexity of the global motion model. Related apparatus, systems, techniques and articles are also described.
Claims
1. A decoder configured to: The receiver receives a bitstream comprising an encoded image having a first region and a second region. The first region has global motion and includes a first consecutive plurality of encoded blocks, and the second region has local motion and includes a second consecutive plurality of encoded blocks. The first region is encoded by a motion model that encodes all blocks of the first region in increasing order of complexity. The motion model is one of translational motion, four-parameter affine motion, or six-parameter affine motion. The complexity of the motion model is a function of the number of motion vectors required to inter-frame encode the blocks using the motion model. A motion model is used to decode the encoded image for each block of the encoded image, which includes the first consecutive plurality of encoded blocks and the second consecutive plurality of encoded blocks, wherein the complexity of the motion model for each block is no greater than the complexity of the first region of the global motion; The decoded image is stored in a buffer.
2. A method for transmitting an encoded bitstream, the encoded bitstream being decoded by a decoder receiving the bitstream, the method comprising: Receive input video signals; Generate an encoded bitstream, the encoded bitstream including an encoded image, the encoded image having a first region and a second region, the first region having global motion and including a first consecutive plurality of encoded blocks, the second region having local motion and including a second consecutive plurality of encoded blocks, the first region being encoded by a motion model in ascending order of complexity, the motion model being one of translational motion, four-parameter affine motion, or six-parameter affine motion, the complexity of the motion model being a function of the number of multiple motion vectors required to inter-frame encode the blocks using the motion model; The encoded bitstream is transmitted through a channel to a decoder, wherein the decoder has instructions and is configured to: A motion model is used to decode the encoded image for each block of the encoded image, which includes the first consecutive plurality of encoded blocks and the second consecutive plurality of encoded blocks, wherein the complexity of the motion model for each block is no greater than the complexity of the first region of the global motion; The decoded image is stored in a buffer.
3. A video encoder, comprising circuitry configured to: Receive input video signals; A encoded bitstream is generated, comprising an encoded image having a first region and a second region. The first region has global motion and includes a first series of consecutive encoded blocks, and the second region has local motion and includes a second series of consecutive encoded blocks. The first region is encoded by a motion model that encodes all blocks of the first region in increasing order of complexity. The motion model is one of translational motion, four-parameter affine motion, or six-parameter affine motion. The complexity of the motion model is a function of the number of motion vectors required to inter-frame encode the blocks using the motion model. The encoded bitstream is decoded by a method comprising: A motion model is used to decode the encoded image for each block of the encoded image, which includes the first consecutive plurality of encoded blocks and the second consecutive plurality of encoded blocks, wherein the complexity of the motion model for each block is no greater than the complexity of the first region of the global motion; The decoded image is stored in a buffer.