Video Motion Prediction with Absolute-Location Lookup Tables
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video coding methods have low coding efficiency, particularly in handling video frames with text content, due to inefficient motion vector encoding and decoding processes.
Innovation Solution
A video processing method that predicts motion information of a current block based on an absolute location in a lookup table, utilizing candidate locations within the current video frame, especially for repeated blocks, to improve motion vector encoding/decoding efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional motion vector encoding methods are used, then all motion information must be explicitly coded, but this increases coding complexity and bit consumption
Solution Approach 1:
The patent uses a lookup table to store and reuse motion information of previously encoded blocks. When a repeated block is detected, its motion vector is copied from the lookup table entry corresponding to its absolute location, avoiding redundant encoding while maintaining accuracy
Solution Approach 2:
The lookup table is pre-populated with motion information of blocks at specific absolute locations during the encoding process. This preliminary storage enables rapid retrieval and copying of motion vectors for repeated blocks without real-time computation
2Reliability
If motion information of all blocks is explicitly coded, then complete motion data is achieved, but the number of coding bits increases
Solution Approach 1:
Instead of encoding motion vectors for all blocks, the patent copies motion information from the lookup table for repeated blocks. The lookup table stores motion vectors at absolute locations, and repeated blocks retrieve their motion data through copying, significantly reducing bit consumption while maintaining completeness
Solution Approach 2:
The patent changes the encoding parameter from explicit motion vector coding to indirect reference coding. By using absolute locations as keys in the lookup table, the system transitions from coding full motion information to coding only location references, reducing the quantity of coded data
Data Source
AI summary
A video processing method related to the field of video encoding and decoding, includes a coder side that predicts motion information of a current block based on an absolute location, in a lookup table, of a block in a current video frame.