Implicit Multi-Pass Decoder Motion Vector Refinement
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video coding standards like HEVC and VVC face challenges in optimizing motion vector refinement for improved compression efficiency and quality, particularly in handling bi-predictive blocks with explicit or implicit motion parameter signaling.
Innovation Solution
Implementing a multi-pass decoder-side motion vector refinement (MP-DMVR) system that uses implicit signaling to refine motion vectors based on bilateral matching and bi-directional optical flow, adapting to different modes to minimize costs and improve prediction accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If explicit syntax signaling is used to transmit motion parameters, then motion vector precision is improved, but bitstream overhead increases
Solution Approach 1:
The decoder performs motion vector refinement autonomously using the received motion vector difference and reference picture data without requiring additional explicit signaling from the encoder. The decoder calculates refined motion vectors by comparing the current block with reference blocks using bilateral matching or optical flow methods, making the system self-sufficient for achieving high precision.
Solution Approach 2:
A motion vector difference is introduced as an intermediary parameter that carries essential refinement information in a compact form. This intermediary enables the decoder to derive multiple refined motion vectors for different prediction directions without transmitting all detailed motion parameters explicitly, thus reducing overhead while maintaining precision.
2Measurement precision
If multi-pass refinement is applied to improve prediction accuracy, then picture quality is improved, but computational complexity increases
Solution Approach 1:
The motion vector refinement process is divided into multiple passes, each targeting different prediction directions (e.g., list 0 and list 1). Each pass refines motion vectors independently using the same efficient algorithm, allowing the complex task to be broken down into manageable segments that can be processed systematically with controlled computational cost.
Solution Approach 2:
The system applies refinement selectively based on the prediction mode and block characteristics. For bi-predictive blocks, refinement is applied to both prediction lists, while for other modes, refinement may be applied to a single list or omitted entirely. This partial application of refinement reduces unnecessary computational complexity while maintaining accuracy where it matters most.
3Productivity
If bilateral matching and optical flow methods are used for motion vector refinement, then compression efficiency is improved, but processing time increases
Solution Approach 1:
The system performs preliminary motion vector estimation using standard inter prediction methods before applying bilateral matching or optical flow refinement. This preliminary estimation provides a good initial guess that reduces the search space and number of iterations required for refinement, thereby decreasing processing time while maintaining compression efficiency gains.
Solution Approach 2:
The refinement methods are applied selectively to different regions and blocks based on their characteristics. Blocks with high motion complexity or those that benefit most from refinement undergo bilateral matching or optical flow processing, while simpler blocks use faster refinement methods or skip refinement entirely. This local differentiation optimizes the balance between compression efficiency and processing time.
Data Source
AI summary
A video coding system that uses implicit signaling for multiple-pass decoder-side motion vector refinement (MP-DMVR) is provided. A video coder receives data for a block of pixels to be encoded or decoded as a current block of a current picture of a video. The current block is associated with a first motion vector referring a first initial predictor and a second motion vector referring a second initial predictor. The video coder refines the first and second motion vectors to minimize first, second, and third costs according to first, second, and third refinement modes, respectively. The video coder selects a refinement mode based on a comparison of the first, second, and third costs. The video coder encodes or decodes the current block by using the selected refinement mode to modify the first and second motion vectors to reconstruct the current block.