Arbitrary Partition Transform for Video Signal Compression
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current video signal processing methods face challenges in efficiently handling next-generation video content with high spatial resolution and high frame rates, requiring a balance between low complexity and high compression efficiency, especially as existing separable transforms have low performance in signal compression.
Innovation Solution
The method employs an arbitrary partition transform, which involves determining an optimal partition set and transform matrix set using an optimization function to perform a multi-dimensional arbitrary partition transform, allowing for efficient compression coding by approximating non-separable transforms using one-dimensional transforms and supporting non-square blocks and multiple dimensions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If separable transform is used, then computational complexity is reduced to about N³, but compression efficiency deteriorates compared to non-separable transform
Solution Approach 1:
The patent segments the two-dimensional transform into multiple one-dimensional transforms applied sequentially on different partitions. By dividing the N×N block into smaller partitions and applying 1D transforms along specific directions (rows, columns, or diagonals), the complexity is reduced from O(N⁴) to O(N³) while maintaining better compression efficiency through adaptive partition selection and directional transform matching.
Solution Approach 2:
The patent introduces dynamic adaptability by allowing the transform direction and partition size to vary based on the characteristics of the input data. The encoder can dynamically select between horizontal, vertical, or diagonal transform directions and adjust partition dimensions to match the local signal characteristics, thereby optimizing compression efficiency while maintaining computational feasibility.
2Loss of information
If non-separable transform is used, then compression efficiency is improved, but computational complexity increases to N⁴
Solution Approach 1:
The patent decomposes the computationally intensive non-separable transform into a sequence of simpler one-dimensional transforms applied on segmented partitions. This segmentation approach maintains the compression efficiency benefits of non-separable transforms by capturing directional signal characteristics while reducing the computational burden from O(N⁴) to O(N³) through the use of multiple 1D transform stages.
Solution Approach 2:
The patent replaces the direct application of complex non-separable transform mechanics with an alternative mechanical system consisting of multiple sequential one-dimensional transforms. This substitution achieves similar compression performance through a different computational mechanism that is more efficient and easier to implement in practice.
3Loss of information
If transform is applied to high spatial resolution video content, then compression efficiency is improved, but processing power requirement increases tremendously
Solution Approach 1:
The patent segments large high-resolution video blocks into smaller partitions, allowing 1D transforms to be applied independently on each partition. This segmentation reduces the processing power requirement by avoiding the O(N⁴) complexity of applying full 2D transforms on large blocks, while still achieving good compression efficiency through adaptive partitioning and directional transform selection.
Solution Approach 2:
The patent introduces dynamic adaptability in transform application, where the encoder can selectively apply transforms only to partitions that benefit from them based on local signal characteristics. This dynamic approach reduces overall processing power consumption by avoiding unnecessary transforms on already-compressed or simple regions while maintaining high compression efficiency on complex regions.
4Adaptability or versatility
If arbitrary partition transform is applied, then compression efficiency and adaptability are improved, but determination of optimal partition set increases complexity
Solution Approach 1:
The patent applies preliminary action by pre-defining a set of candidate partition configurations and transform directions before processing the actual video data. These predefined patterns serve as templates that guide the partitioning process, reducing the complexity of optimal partition determination while maintaining adaptability through selective application to different block types and content characteristics.
Data Source
AI summary
Provided is a method for processing a video signal on the basis of an arbitrary partition transform, the method comprising the steps of: determining a partition set for input data, the partition set indicating a set of partitions divided from a data block on the basis of a scanning pattern; determining a transform matrix set corresponding to the partition set; obtaining an arbitrary partition transform vector for each partition by applying the transform matrix set to the partition set; and obtaining the arbitrary partition transform coefficient for the input data by inverse-mapping the arbitrary partition transform vector.