Intra Prediction Mode Derivation for Lower Video Coding Overhead
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The increasing demand for high-resolution, high-quality video and immersive media formats necessitates more efficient image compression techniques to reduce transmission and storage costs while maintaining or improving prediction accuracy.
Innovation Solution
The implementation of Template-based Inter Mode Derivation (TIMD) and Decoder Side Intra Mode Derivation (DIMD) modes for intra prediction, which derive prediction samples based on block templates and surrounding reference samples, respectively, to enhance coding efficiency and reduce overhead in signaling intra prediction modes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If conventional intra prediction methods are used, then implementation is simple, but coding efficiency is insufficient for high-resolution video
Solution Approach 1:
The decoder derives the intra prediction mode independently using template matching and gradient calculations on reference samples, without requiring explicit mode signaling from the encoder. This self-service approach reduces overhead and improves coding efficiency while maintaining implementation feasibility through standardized algorithms.
Solution Approach 2:
The patent changes the parameter representation from explicit mode indices to derived modes based on gradient directions and template matching results. By transforming the prediction mode selection into a continuous gradient-based derivation process, the system achieves better adaptation to local image characteristics and improved coding efficiency.
2Measurement precision
If more bits are transmitted for high-resolution video, then video quality is improved, but transmission and storage costs increase
Solution Approach 1:
The encoder and decoder use feedback from gradient calculations and template matching to adaptively select prediction modes that best match the local image characteristics. This feedback mechanism enables more accurate prediction, reducing the residual information that needs to be transmitted and thereby lowering transmission and storage costs while maintaining high video quality.
Solution Approach 2:
The system performs preliminary gradient calculations and template matching on reference samples before actual prediction, preparing optimal prediction modes in advance. This preliminary action enables more accurate prediction for high-resolution video, reducing the amount of data that needs to be transmitted and stored while maintaining superior video quality.
3Loss of information
If TIMD or DIMD mode is applied, then overhead in signaling intra prediction mode is reduced, but calculation complexity increases
Solution Approach 1:
The patent replaces the mechanical signaling system with an algorithmic derivation system. Instead of transmitting explicit mode indices through the bitstream, the system uses gradient calculations and template matching algorithms to derive prediction modes at the decoder side, eliminating signaling overhead while managing calculation complexity through efficient algorithm design.
Data Source
AI summary
According to one embodiment of the present disclosure, a method by which a decoding device decodes an image is provided. The method comprises the steps of: acquiring image information from a bit-stream; determining whether at least one from among a template-based inter mode derivation (TIMD) mode and decoder-side intra mode derivation (DIMD) mode is applied to the current block; generating a prediction sample of the current block on the basis of the at least one from among the TIMD mode and DIMD mode; and generating a reconstructed sample of the current block on the basis of the prediction sample of the current block, wherein the TIMD mode is derived from a template of the current block and the DIMD mode is derived on the basis of the gradient of adjacent reference samples of the current block.