Adaptive Perceptual Transfer Functions for HDR Video Quantization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video encoding and decoding schemes struggle to effectively handle High Dynamic Range (HDR) and Wide Color Gamut (WCG) video sequences, as they are not designed to directly process HDR or WCG data, leading to quantization distortion and loss of detail due to fixed coding transfer functions that do not account for content characteristics.
Innovation Solution
Adaptive pre-processing and post-processing methods are applied to HDR and WCG video data before encoding and after decoding, using color space conversions and perceptual transfer functions to redistribute data, allowing for efficient compression and reconstruction of HDR and WCG data, with metadata passed between encoder and decoder to ensure accurate inversion of processing steps.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If a fixed coding transfer function is used to convert linear input values into non-linear values, then the conversion can be performed efficiently with standard compression algorithms, but quantization distortion is introduced because the fixed mapping does not account for content characteristics
Solution Approach 1:
The patent applies a perceptual transfer function that dynamically adapts to the content characteristics of the input video sequence. Instead of using a fixed coding transfer function, the system analyzes the input content and adjusts the transfer function parameters accordingly, allowing the mapping between linear and non-linear values to change based on the actual video data. This dynamic adaptation enables optimal quantization allocation for each specific content while maintaining compatibility with standard compression algorithms.
2Device complexity
If 16-bit half-precision floating point color values are directly quantized to 10-bit values, then the bit depth is reduced for compression compatibility, but video resolution is substantially reduced and significant distortion is introduced
Solution Approach 1:
The patent changes the parameter of the transfer function to optimize the quantization process. By applying a perceptual transfer function that maps a subset of the HDR range to the limited SDR range, the system achieves more efficient use of the available quantization levels. This parameter transformation allows the system to maintain color precision and detail in the visible range while reducing the overall dynamic range to be compatible with standard 10-bit compression algorithms.
3Productivity
If a fixed coding transfer function maps the full HDR range to SDR values, then the compression algorithm can process the data, but poor allocation of quantization levels results in contouring and banding artifacts
Solution Approach 1:
The patent applies local quality by analyzing the specific content characteristics of each input video sequence and adjusting the perceptual transfer function accordingly. Instead of using a one-size-fits-all fixed transfer function, the system adapts the mapping parameters to match the actual luminance and color distribution in the video content. This localized optimization ensures that quantization levels are allocated according to the specific needs of each scene, reducing contouring and banding artifacts while maintaining encoding efficiency.
Data Source
AI summary
A method of encoding a digital video data applies adaptive pre-processing to data representing high dynamic range (HDR) and/or wide color gamut (WCG) image data prior to encoding and complementary post-processing to the data after decoding in order to allow at least partial reproduction of the HDR and/or WCG data. The example methods apply one or more color space conversions, and a perceptual transfer functions to the data prior to quantization. The example methods apply inverse perceptual transfer functions and inverse color space conversions after decoding to recover the HDR and/or WCG data. The transfer functions are adaptive so that different transfer functions may be applied to video data sets including different groups of frames, frames or processing windows in a single frame. Information on the data set and information on the applied transfer function is passed as metadata from the encoder to the decoder.