Method, an apparatus and a computer program product for video encoding and video decoding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video coding systems face challenges in efficiently compressing video data due to the high computational complexity of Karhunen-Loeve transform (KLT) and the need for practical implementations of secondary transforms in video codecs, which are hindered by the complexity of calculating eigenvectors from large covariance matrices.
Innovation Solution
Implementing a herd-KLT approach that splits secondary transforms into horizontal and vertical processing, with independent transform matrices and statistical models, allowing for adaptive updates based on statistical information to reduce computational complexity and improve efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If Karhunen-Loeve transform is used for video coding, then compression efficiency is improved, but computational complexity increases significantly
Solution Approach 1:
The patent divides the full KLT into multiple secondary transforms (ST0, ST1, ST2, etc.), each handling specific subsets of transform coefficients. This segmentation allows the complex KLT to be implemented through simpler, manageable transform stages that can be processed sequentially, reducing the computational burden while maintaining compression efficiency.
Solution Approach 2:
The patent introduces a parameter N to control the number of secondary transform stages applied. By adjusting N, the system can adapt the computational complexity to match available resources while maintaining adequate compression performance. This parameterized approach allows flexible trade-off between compression efficiency and computational load.
2Productivity
If secondary transforms are implemented in video codecs, then coding performance is improved, but implementation complexity increases due to eigenvector calculation
Solution Approach 1:
The patent pre-calculates and stores transform matrices for multiple secondary transforms before actual video encoding/decoding. These pre-computed matrices eliminate the need for real-time eigenvector calculations during video processing, significantly reducing implementation complexity while preserving the performance benefits of KLT-based secondary transforms.
Solution Approach 2:
The patent uses pre-computed transform matrices as templates that are copied and applied repeatedly during video coding. Instead of calculating eigenvectors for each video block, the system copies and applies the appropriate pre-computed matrix, dramatically simplifying the implementation while maintaining coding performance.
Data Source
AI summary
The embodiments relate to a method and technical equipment for implementing the method. The method comprises receiving a block of input samples; performing at least a first inverse transform on a first set of input samples, resulting in a first set of inverse transformed samples; performing at least a second inverse transform on a second set of input samples, resulting in a second set of inverse transformed samples; determining statistical information related to the first set of inverse transformed samples and for determining statistical information related to the second set of inverse transformed samples; updating the first inverse transform based on the determined statistical information related to the first set of inverse transformed samples; and updating the second inverse transform based on the determined statistical information related to the second set of inverse transformed samples.