Deterministic Bounding Box for Decoder Motion Vector Derivation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video coding technologies face challenges in efficiently deriving motion vectors at the decoder side, leading to increased decoding latency and complexity due to hardware implementation issues such as spatial motion vector propagation and neighboring reconstruction sample accessing.
Innovation Solution
The proposed solution involves determining a deterministic bounding box to retrieve reference samples for decoder-side motion vector derivation (DMVD), which limits the search range and reduces the number of reference samples fetched, thereby simplifying the hardware implementation and reducing decoding latency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If decoder-side motion vector derivation (DMVD) is performed using existing video coding technologies, then motion compensation accuracy is improved, but decoding latency and hardware complexity increase
Solution Approach 1:
The patent segments the reference picture into multiple regions including a deterministic bounding box and other regions. The deterministic bounding box is defined based on the motion vector candidate position and a predetermined offset, creating a focused search area. This segmentation allows the decoder to concentrate computational resources on the most relevant region, improving motion compensation accuracy while reducing overall decoding latency by avoiding exhaustive search across the entire reference picture.
Solution Approach 2:
The patent applies local quality by treating different regions of the reference picture differently. The deterministic bounding box receives priority processing with guaranteed access, while other regions have access constrained by rate control parameters. This local differentiation optimizes the balance between motion compensation accuracy (by ensuring adequate sampling in critical regions) and decoding efficiency (by limiting access to less critical regions), thereby reducing hardware complexity and decoding latency.
2Measurement precision
If decoder-side motion vector derivation (DMVD) is performed using existing video coding technologies, then motion compensation accuracy is improved, but hardware complexity increases
Solution Approach 1:
The patent segments the reference picture into a deterministic bounding box and other regions, with the bounding box size determined by a predetermined offset from the motion vector candidate position. This segmentation simplifies hardware implementation by creating a fixed, predictable access pattern for the most critical region, reducing the complexity of memory access control logic while maintaining motion compensation accuracy through focused sampling in the bounding box area.
Solution Approach 2:
The patent performs preliminary action by pre-defining the deterministic bounding box parameters (position and size based on predetermined offset) before the actual motion compensation process. This allows the hardware to be configured with fixed access patterns and memory addresses in advance, reducing runtime complexity and simplifying the hardware architecture. The rate control parameter is also determined in advance, further reducing runtime decision-making complexity.
3Measurement precision
If a large search range is used for retrieving reference samples in DMVD, then motion compensation accuracy is improved, but decoding complexity and latency increase
Solution Approach 1:
The patent applies local quality by differentiating between the deterministic bounding box region and other regions in the reference picture. The bounding box, positioned around the motion vector candidate with a predetermined offset, receives enhanced sampling density and guaranteed access. Other regions have access controlled by rate control parameters. This local quality differentiation maintains motion compensation accuracy by ensuring adequate sampling where most needed, while improving decoding efficiency by limiting access to less critical regions, thus resolving the trade-off between accuracy and productivity.
Data Source
AI summary
A device for decoding video data includes a memory configured to store video data; and one or more processors implemented in circuitry and configured to: determine a deterministic bounding box from which to retrieve reference samples of reference pictures of video data for performing decoder-side motion vector derivation (DMVD) for a current block of the video data; derive a motion vector for the current block according to DMVD using the reference samples within the deterministic bounding box; form a prediction block using the motion vector; and decode the current block using the prediction block.