Tensor-Product B-Spline Prediction for Fast Mobile HDR Decoding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Reconstructing HDR video on mobile devices is challenging due to limited computing power, storage, and battery life, and existing legacy schemes fail to provide satisfactory static mapping for wide color gamut, especially close to ITU-R Rec. BT. 2020, without hardware accelerators.
Innovation Solution
Implementing a tensor-product B-spline (TPB) predictor with a 3D look-up table (LUT) for efficient SDR-to-HDR mapping, using TPB metadata to reduce computational complexity and support backward compatibility, and combining it with legacy methods in a mobile-optimized profile.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If tensor-product B-spline predictor is implemented for HDR reconstruction, then HDR reconstruction quality is improved, but computational complexity increases
Solution Approach 1:
The patent pre-calculates and stores B-spline basis functions and their derivatives in lookup tables before runtime. This preliminary computation eliminates the need for complex real-time calculations during HDR reconstruction, allowing mobile devices to achieve high-quality results with minimal computational overhead.
Solution Approach 2:
The patent uses lookup tables that store pre-computed basis function values and derivatives, effectively copying results of complex mathematical operations into accessible memory structures. This allows the system to retrieve pre-calculated values instead of performing computationally intensive operations during actual HDR reconstruction.
2Measurement precision
If tensor-product B-spline predictor with 3D LUT is used for SDR-to-HDR mapping, then mapping accuracy is improved, but decoding time increases
Solution Approach 1:
The patent pre-computes the 3D lookup table containing transformed coordinate values and basis function derivatives during an initialization phase. This preliminary action allows the decoding process to simply retrieve pre-calculated values rather than performing complex transformations in real-time, thus improving decoding speed while maintaining accuracy.
Solution Approach 2:
The patent divides the complex SDR-to-HDR mapping process into separate, pre-computed components stored in the 3D LUT. By segmenting the transformation into discrete lookup operations rather than continuous mathematical evaluations, the system achieves both accuracy and computational efficiency.
3Adaptability or versatility
If legacy prediction schemes are used for HDR reconstruction, then device compatibility is maintained, but reconstruction quality deteriorates
Solution Approach 1:
The patent designs the TPB predictor to work with standard video coding formats and interfaces, making it universally compatible with existing HDR workflows and devices. The predictor can be integrated into current decoding pipelines without requiring specialized hardware, thus maintaining compatibility while improving quality.
Data Source
AI summary
Tensor-Product B-splines (TPB) have been shown to improve video quality when used to represent reshaping functions to map reshaped standard dynamic range content into high dynamic range (HDR) content; however, TPB prediction is computationally intensive and may not be supported by legacy devices. Methods and systems for backwards-compatible signaling of TPB-related metadata and a fast TPB prediction method are presented to overcome both of these limitations. Computation overhead for a TPB-based 3D look-up table is reduced by using temporary two-dimensional arrays. A remapping of the most significant bits of a legacy bit-depth parameter allows for backwards compatibility.