DIMD Mode-Based Intra Prediction for Lower-Signaling Image Decoding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The increasing demand for high-resolution and high-quality images/video, particularly in immersive media like VR and AR, leads to higher data transmission and storage costs, necessitating a more efficient image coding technique to compress and transmit these formats effectively.
Innovation Solution
The method involves deriving intra prediction modes using Decoder-side Intra Mode Derivation (DIMD) and Template-based Intra Mode Derivation (TIMD) information, parsing flags for MIP, DIMD, and TIMD to optimize image coding, and generating prediction and reconstructed samples based on these modes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If conventional intra prediction mode signaling is used, then image quality can be maintained, but the amount of data transmitted and storage costs increase
Solution Approach 1:
The patent extracts only the essential information needed for intra prediction mode derivation by using flags (dimd_flag, timd_flag) to indicate whether DIMD or TIMD modes are applied, rather than transmitting complete mode information. This reduces the signaling data amount while maintaining the ability to reconstruct image quality.
Solution Approach 2:
The patent enables the decoder to derive intra prediction modes autonomously using DIMD (Decoder-side Intra Mode Derivation) and TIMD (Template-based Intra Mode Derivation) techniques. The decoder uses local template matching and gradient calculations to determine prediction modes without requiring extensive signaling from the encoder, thereby reducing transmitted data while maintaining image quality.
2Productivity
If DIMD and TIMD modes are implemented, then compression efficiency improves, but device complexity increases
Solution Approach 1:
The patent segments the intra prediction process into distinct modes (DIMD, TIMD, and conventional modes) that can be selectively applied. Each mode handles specific types of image content differently, allowing the system to achieve high compression efficiency for suitable content while maintaining simplicity for other cases through flag-based mode selection.
Solution Approach 2:
The patent applies complex DIMD and TIMD processing only when beneficial, as indicated by the dimd_flag and timd_flag. Rather than always applying the most complex methods, the system uses partial action by selectively enabling these modes only when they improve compression efficiency, thereby balancing complexity and performance.
Data Source
AI summary
Provided is an image decoding method performed by a decoding apparatus, the image decoding method including receiving image information including at least one of decoder-side intra mode derivation (DIMD)-related information or template-based intra mode derivation (TIMD)-related information from a bitstream, deriving an intra prediction mode of a current block on the basis of the at least one of the DIMD-related information or the TIMD-related information, generating prediction samples of the current block on the basis of the intra prediction mode, and generating reconstructed samples of the current block on the basis of the prediction samples of the current block. The image information includes matrix-based intra prediction (MIP) flag information, the DIMD-related information includes DIMD flag information indicating whether a DIMD mode is applied to the current block, the TIMD-related information includes TIMD flag information indicating whether a TIMD mode is applied to the current block, and the DIMD flag information or the TIMD flag information is parsed after the MIP flag information.