Transform Kernel Sharing for Video Coding

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing video coding technologies face challenges in efficiently reducing redundancy in video signals while maintaining acceptable distortion levels, particularly in high-resolution and high-frame-rate video streams.

Innovation Solution

The proposed solution involves a method for transform kernel sharing in video encoding and decoding, where a set of transform kernels with shared high-frequency basis vectors and individualized low-frequency basis vectors are used to process video data blocks.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Loss of energy

If traditional video coding techniques are used to compress high-resolution high-frame-rate video, then bandwidth and storage requirements are reduced, but distortion between original and reconstructed signals increases

Engineering Contradiction:
Improvebandwidth and storage requirementsVSAvoiddistortion between original and reconstructed signals
Core Design Contradiction:
Loss of energyVSLoss of information

Solution Approach 1:

The video signal is segmented into multiple blocks that are processed independently through transform coding. Each block is transformed into frequency domain coefficients, allowing selective compression of different frequency components while maintaining overall signal fidelity within acceptable distortion limits

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The transform kernel parameters are optimized to achieve better energy compaction in the frequency domain. By adjusting transform parameters and using adaptive quantization, the coding system achieves higher compression ratios while controlling the distortion between original and reconstructed signals

Inventive Principle:
Principle #35Parameter changes

2Productivity

If multiple transform kernels are used to improve compression performance, then coding efficiency increases, but memory requirements for storing transform kernels increase

Engineering Contradiction:
Improvecoding efficiencyVSAvoidmemory requirements for transform kernels
Core Design Contradiction:
ProductivityVSQuantity of substance

Solution Approach 1:

Multiple transform kernels are merged into a single unified transform kernel through adaptive weighting and combination techniques. The system dynamically selects and combines basis functions from multiple kernels to achieve compression performance comparable to using multiple separate kernels, while storing only one combined kernel in memory

Inventive Principle:
Principle #5Merging (Combining)

Solution Approach 2:

A single transform kernel is designed to serve multiple functions by incorporating adaptive parameters that allow it to behave like different specialized kernels depending on the input block characteristics. This universal kernel can adapt to various video content types (intra-picture, inter-picture, different frequencies) without requiring separate dedicated kernels for each case

Inventive Principle:
Principle #6Universality (Multi-functionality)

Data Source

PatentUS12335525B2Orthogonal transform generation with subspace constraint
Publication Date: 2025.06.17 TENCENT AMERICA LLC
  • US12335525B2 patent drawing
  • US12335525B2 patent drawing
  • US12335525B2 patent drawing

AI summary

This disclosure relates to a transform kernel sharing in video encoding and decoding. For example, a method is disclosed for such transform kernel sharing. The method may include identifying a plurality of transform kernels, wherein each of the plurality of transform kernels comprises a set of basis vectors from low to high frequencies; N high-frequency basis vectors of two or more of the plurality of transform kernels are shared, N being a positive integer; and low-frequency basis vectors of the two or more of the plurality of the transform kernels other than the N high-frequency basis vectors are individualized. The method may further include extracting a data block from a video bitstream; selecting a transform kernel from the plurality of transform kernels based on information associated with the data block; and applying the transform kernel to at least a portion of the data block to generate a transformed block.