Hybrid Transform Kernel Selection for Video Decoding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current video compression technologies, such as those beyond High Efficiency Video Coding (HEVC), face challenges in efficiently compressing video data due to limitations in inter-prediction and intra-prediction methods, particularly in handling large block sizes and diverse prediction modes.
Innovation Solution
The proposed solution involves a method for decoding image data that selects a hybrid transform kernel using either an implicit or explicit method. This method considers associations between prediction modes and block sizes to choose from available hybrid transform kernels, which include combinations of vertical and horizontal transform types, such as Line Graph Transforms (LGT), to perform inverse transform coding.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If traditional inter-prediction and intra-prediction methods are used for video compression, then the compression process is simple, but compression efficiency is insufficient especially for large block sizes and diverse prediction modes
Solution Approach 1:
The patent segments the prediction process into multiple modes (intra-prediction, inter-prediction, bi-prediction) and further divides blocks into smaller units (e.g., 4x4, 8x8, 16x16, 32x32, 64x64) that can be processed independently. This segmentation allows the system to handle large blocks by dividing them into manageable smaller blocks, improving compression efficiency without overwhelming complexity
Solution Approach 2:
The patent implements dynamic prediction methods where the choice between intra-prediction, inter-prediction, or bi-prediction modes is made adaptively based on the specific block characteristics and context. The system dynamically adjusts prediction parameters and modes during encoding/decoding, allowing optimization for each specific case rather than using a fixed approach
2Productivity
If hybrid transform kernels are used to improve decoding performance, then compression efficiency increases, but the complexity of selecting appropriate kernels increases
Solution Approach 1:
The patent uses parameter changes by varying transform kernel types (e.g., DCT, DST, Walsh-Hadamard, identity) and their corresponding parameters (block size, prediction mode) to optimize decoding performance. The system changes these parameters adaptively based on the input data characteristics, achieving high performance without requiring complex manual selection
Solution Approach 2:
The transform kernel selection process is made self-service through automated rules that map prediction modes and block sizes to appropriate kernels. The system self-determines the best kernel based on pre-established associations between prediction modes, block dimensions, and transform types, eliminating the need for complex external selection mechanisms
3Speed
If larger block sizes are processed to reduce computational complexity, then processing speed increases, but compression efficiency decreases
Solution Approach 1:
The patent segments large blocks into smaller blocks (4x4, 8x8, 16x16, 32x32, 64x64) that can be processed independently. This segmentation allows the system to maintain high processing speed by working with smaller units while achieving better compression efficiency through more precise processing of each segment
Solution Approach 2:
The patent applies partial action by processing only the necessary portions of each block using appropriate transform kernels. Instead of uniformly processing entire large blocks, the system applies selective transforms based on local characteristics, achieving efficient compression without excessive computational overhead
Data Source
AI summary
A method of decoding image data is provided. The method may include generating a dequantized block of an image based on a coded bitstream; determining whether to use one from among an implicit method and an explicit method for selecting a hybrid transform kernel from among one or more hybrid transform kernel that are available for decoding the dequantized block, wherein the one or more hybrid transform kernel are available for decoding the dequantized block based on associations between prediction modes and sizes of blocks; selecting, by using the one from among the implicit method and the explicit method, the hybrid transform kernel from among the one or more hybrid transform kernel; and performing inverse transform coding of the dequantized block based on the selected hybrid transform kernel.