Video Encoding Orthogonal Transform Complexity Reduction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The existing mode-dependent directional transform (MDDT) in advanced video coding (AVC) faces challenges with high calculation complexity, especially for large block sizes, leading to increased circuit scale and manufacturing costs for hardware implementation and processing time for software implementation, due to the use of separable-type Karhunen-Loeve transform (KLT).
Innovation Solution
The proposed solution involves a two-stage orthogonal transform process, where a discrete wavelet transform (DWT) with lower calculation complexity is performed first, followed by a separable-type KLT only on the low-frequency components, reducing the overall calculation burden and enabling more efficient encoding and decoding.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If separable-type KLT is used for orthogonal transform processing, then encoding efficiency is improved, but calculation complexity increases significantly for large block sizes
Solution Approach 1:
The patent segments the orthogonal transform processing into two distinct stages: first performing a separable-type KLT on the entire image block, then performing a non-separable type KLT only on the low-frequency components. This segmentation allows the computationally intensive non-separable KLT to be applied only where necessary (low-frequency components), thereby reducing overall calculation complexity while maintaining encoding efficiency for the most important frequency components.
Solution Approach 2:
The patent applies partial action by performing the non-separable type KLT only on the low-frequency components rather than on the entire image block. This partial application of the more complex transform to the most significant components achieves a balance between encoding efficiency and calculation complexity, avoiding the excessive computation that would result from applying the non-separable KLT to all frequency components.
2Productivity
If separable-type KLT is implemented in hardware, then encoding can be performed, but circuit scale and manufacturing cost increase
Solution Approach 1:
The patent segments the transform processing to perform separable-type KLT first (which has lower hardware complexity) and then non-separable type KLT only on low-frequency components. This segmentation reduces the overall hardware requirements compared to implementing a full non-separable KLT for the entire block, thereby reducing circuit scale and manufacturing cost while maintaining encoding capability.
Solution Approach 2:
The patent applies partial action by implementing the more complex non-separable KLT only for low-frequency components in the hardware architecture. This partial implementation reduces the total number of required circuit elements and computational resources, thereby lowering manufacturing cost while still providing effective encoding for the most important frequency components.
3Productivity
If separable-type KLT is implemented in software, then encoding can be performed, but processing time increases and real-time reproduction becomes difficult
Solution Approach 1:
The patent segments the software processing into two stages: first performing the separable-type KLT (which is computationally lighter), then performing the non-separable type KLT only on low-frequency components. This segmentation significantly reduces the total processing time compared to applying the non-separable KLT to the entire block, thereby enabling real-time reproduction while maintaining encoding capability.
Solution Approach 2:
The patent applies partial action by performing the more computationally intensive non-separable KLT only on low-frequency components rather than the entire block. This partial application reduces the total number of operations required, thereby decreasing processing time and enabling real-time software implementation while still providing effective encoding.
Data Source
AI summary
The present technology relates to an encoding device and an encoding method, and a decoding device and a decoding method that are capable of reducing the calculation amount of orthogonal transform processing or inverse orthogonal transform processing. A DWT unit (91) performs a DWT of which the calculation amount is smaller than that of a KLT for residual information. KLT units (92-0 to 92-8) perform separable-type KLTs for a low-frequency component of the residual information that is acquired as a result of the DWT using bases of KLTs of the corresponding intra prediction modes. A coefficient acquired as a result of the KLT and a high-frequency component of the residual information that is acquired as a result of the DWT are losslessly encoded. The present technology can be applied to, for example, an image encoding device.