Inter Prediction Using POC LSB for Long-Term Reference Indexing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional video codecs face challenges in efficiently encoding and decoding high-resolution video content due to limitations in inter prediction methods, particularly in utilizing long-term reference images for motion compensation, which require separate image numbers and result in inefficient storage and indexing.
Innovation Solution
The method involves using the least significant bit (LSB) information of the picture order count (POC) as a long-term reference index, allowing for the identification of long-term reference images without separate image numbers, thereby optimizing storage and indexing, and enabling infinite ranges of indexes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If separate image numbers are used to identify long-term reference images, then the identification of reference images is achieved, but the storage space and device complexity increase
Solution Approach 1:
The patent merges the reference image identification function into the existing POC field by reusing its bit structure. Instead of maintaining separate image number fields, the POC value itself serves as the reference image identifier, combining storage and identification functions into a single mechanism, thereby reducing overall system complexity and storage requirements
Solution Approach 2:
The POC field is given multiple functions: it continues to serve its traditional purpose of picture ordering while simultaneously functioning as a reference image identifier. This multi-functionality eliminates the need for dedicated image number fields, reducing device complexity while maintaining identification accuracy
2Productivity
If conventional inter prediction methods are used with limited reference images, then the encoding process is simple, but the encoding efficiency for high-resolution video is insufficient
Solution Approach 1:
The patent introduces dynamic reference image selection based on POC values, allowing the encoding system to adaptively choose from a larger pool of reference images including long-term references. This dynamic approach improves encoding efficiency for high-resolution video by selecting the most appropriate reference images for each prediction task, while the underlying mechanism leverages existing POC structures to minimize additional complexity
Data Source
AI summary
Provided are an inter prediction method and a motion compensation method. The inter prediction method includes: performing inter prediction on a current image by using a long-term reference image stored in a decoded picture buffer; determining residual data and a motion vector of the current image generated via the inter prediction; and determining least significant bit (LSB) information as a long-term reference index indicating the long-term reference image by dividing picture order count (POC) information of the long-term reference image into most significant bit (MSB) information and the LSB information.