Sample Clipping for Prediction Refinement with Optical Flow
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Video coding systems using Prediction Refinement with Optical Flow (PROF) face challenges in bit depth extension due to the refined predictors exceeding 16 bits, which complicates SIMD implementation and reduces processing throughput.
Innovation Solution
The solution involves clipping the prediction offset or refined predictors to specific bit ranges, such as [−8192, 8191] or [−32768, 32767], to maintain compatibility with 16-bit SIMD instructions, thereby preventing bit-depth exceeding and ensuring efficient processing.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If Prediction Refinement with Optical Flow (PROF) is used to improve coding efficiency, then prediction accuracy is improved, but bit depth increases beyond 16 bits which complicates SIMD implementation and reduces processing throughput
Solution Approach 1:
The patent applies parameter changes by clipping the prediction offset to a specific bit range (e.g., 14 bits) rather than allowing it to exceed 16 bits. This parameter constraint ensures that the refined predictor remains within the 16-bit range, maintaining compatibility with efficient SIMD instructions while still providing improved prediction accuracy through PROF
Solution Approach 2:
The patent uses partial action by applying clipping only to the prediction offset component rather than to the entire refined predictor calculation process. This selective clipping maintains sufficient prediction accuracy while preventing bit depth expansion, achieving a balance between PROF benefits and SIMD efficiency
2Measurement precision
If prediction offset is not clipped, then prediction accuracy is maintained, but bit depth exceeds 16 bits requiring extended precision operations that reduce processing speed
Solution Approach 1:
The patent changes the parameter range of the prediction offset by imposing a clipping limit (e.g., ±8192 for 14-bit clipping). This parameter constraint ensures that when the prediction offset is added to the original predictor, the result remains within 16 bits, enabling use of fast SIMD instructions while maintaining adequate prediction accuracy
3Measurement precision
If refined predictor exceeds 16 bits, then prediction precision is improved, but SIMD implementation becomes complex and processing efficiency decreases
Solution Approach 1:
The patent applies parameter changes by constraining the prediction offset to a fixed bit width (e.g., 14 bits) through clipping. This ensures that the refined predictor (original predictor + prediction offset) remains within 16 bits, maintaining simplicity of SIMD implementation while still achieving improved prediction precision through the PROF mechanism
Data Source
AI summary
Methods and apparatus of video coding using sub-block based affine mode are disclosed. According to this method, control-point motion vectors (MVs) associated with the affine mode are determined for a block. A sub-block MV is derived for a target sub-block of the block from the control-point MVs for the block. A prediction offset is determined for a target pixel of the target sub-block using information comprising a pixel MV offset from the sub-block MV for the target pixel according to Prediction Refinement with Optical Flow (PROF). The target pixel of the target sub-block is encoded or decoded using a modified predictor. The modified prediction is generated by clipping the prediction offset to a target range and combining the clipped prediction offset with an original predictor.