Unified Inverse CCP and ACT Transform for Video Coding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The existing video coding techniques, such as those described in HEVC, face inefficiencies due to the increased internal bit depth and complexity of operations when combining inverse CCP and inverse ACT transforms, particularly when handling different bit depths for luma and chroma components.
Innovation Solution
The proposed solution involves combining the inverse CCP and inverse ACT transforms into a single step, where the bit-depth difference between luma and chroma samples is considered to determine the right shift value, reducing the internal bit-depth increment and simplifying the multiplication operations by using a bit shift value calculated as 3 plus the maximum of 0 and the difference between the luma and chroma bit depths.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If inverse CCP and inverse ACT transforms are combined into separate processing steps, then the transform process is more detailed and accurate, but the decoder complexity and internal bit-depth increment increase
Solution Approach 1:
The patent combines the inverse CCP and inverse ACT transforms into a single unified transform step. By merging these two previously separate processing stages into one operation, the patent reduces the number of processing steps, decreases internal bit-depth increment, and simplifies decoder complexity while still achieving effective de-correlation of color components. The unified transform uses a bit shift value calculated as 3 plus the maximum of 0 and the difference between luma and chroma bit depths, which optimizes the processing efficiency.
2Adaptability or versatility
If separate processing of inverse CCP and inverse ACT is used, then each transform can be optimized independently, but the overall processing efficiency decreases
Solution Approach 1:
The patent merges inverse CCP and inverse ACT into a single unified transform operation that processes both color component de-correlation and color space transformation simultaneously. This consolidation eliminates the need for separate processing steps, reducing the overall processing time and improving decoding efficiency. The unified transform achieves this by using a single set of equations that incorporate both transform objectives, with the bit shift value (3 + max(0, luma_bit_depth - chroma_bit_depth)) optimizing the combined operation.
3Measurement precision
If different bit depths are handled separately for luma and chroma, then bit-depth precision is maintained, but the number of operations and complexity increase
Solution Approach 1:
The patent dynamically adjusts the bit shift value based on the difference between luma and chroma bit depths. By calculating the bit shift as 3 + max(0, luma_bit_depth - chroma_bit_depth), the transform adapts to different bit-depth configurations without requiring separate processing paths. This parameter-based adaptation maintains precision for different bit-depth scenarios while significantly reducing operational complexity compared to handling each bit-depth case separately.
Data Source
Figure 1
Figure 2
Figure 3A~3B
AI summary
A video coder transforms, based on a difference between a bit depth of an initial luma sample and a bit depth of initial chroma samples, a set of initial samples to a set of modified samples. The set of initial samples may include the initial luma sample and the initial chroma samples. The initial chroma samples may include an initial Cb sample and an initial Cr sample. The set of modified samples may include a modified luma sample and modified chroma samples. The modified chroma samples may include a modified Cb sample and a modified Cr sample. Additionally, the video coder adds the modified samples to corresponding samples of a predictive block to produce reconstructed samples.