Intra Prediction Transform Matrix Combination for Video Encoding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Implementing orthogonal transformation and inverse orthogonal transformation using individual transform bases for multiple prediction modes in video encoding is challenging due to the need for dedicated hardware and increased memory bandwidth or cache size, especially in hardware and software implementations respectively.
Innovation Solution
An image encoding apparatus that includes an intra-prediction unit, a setting unit, and a transforming unit, which sets and uses a combination of vertical and horizontal transform matrices to perform orthogonal transformations, optimizing the process by using a combination of 1D transform matrices A and B to increase coefficient density based on prediction error tendencies, reducing the need for dedicated hardware and memory requirements.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If individual transform bases are used for each prediction mode, then coding efficiency is improved, but hardware complexity increases due to dedicated hardware requirements
Solution Approach 1:
The patent implements a unified transform processing unit that can handle multiple prediction modes using a single set of transform bases. The system determines the appropriate transform basis based on the prediction mode and applies it through a universal processing path, eliminating the need for separate dedicated hardware for each prediction mode while maintaining the benefits of mode-specific optimization.
Solution Approach 2:
The patent changes the parameters of a single transform processing unit dynamically based on the prediction mode. Instead of having fixed dedicated hardware for each mode, the system adjusts the transform basis selection and processing parameters according to the active prediction mode, allowing one hardware unit to adaptively serve multiple functions.
2Productivity
If individual transform bases are used for each prediction mode, then coding efficiency is improved, but memory bandwidth or cache size increases
Solution Approach 1:
The patent merges the storage requirements for multiple transform bases into a single unified memory structure. Instead of having separate memory allocations for each prediction mode's transform bases, the system consolidates them into one shared memory resource that is accessed based on the current prediction mode, thereby reducing overall memory bandwidth requirements and cache size.
Solution Approach 2:
The patent performs preliminary determination of the appropriate transform basis based on the prediction mode before the actual transform operation. This allows the system to load or select only the necessary transform basis data in advance, avoiding the need to maintain all transform bases in high-speed cache simultaneously, thus reducing memory bandwidth consumption.
Data Source
AI summary
An image encoding apparatus includes a setting unit configured to set a combination of a vertical transform matrix and a horizontal transform matrix corresponding to the target image. The combination includes any of a plurality of transform matrices including a first transform matrix and a second transform matrix which increases a coefficient density compared to the first transform matrix if a one-dimensional orthogonal transformation in a direction orthogonal to a line of a group of reference pixels on at least one line is performed on the prediction error in the intra-prediction mode in which the group of reference pixels is referenced to generate an intra-prediction image.