Dynamic Range Conversion Model for HDR Image Signal Processing

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current image processing technologies face challenges in effectively converting images between different dynamic ranges, leading to suboptimal display quality on various devices.

Innovation Solution

A method and apparatus for processing image signal conversion using a dynamic range conversion model (L′=F(L)=a×(p×Ln(k1×p-k2)×Ln+k3)m+b) to improve the quality of conversion between images in different dynamic ranges, adapting to display devices for better visual output.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Ease of manufacture

If linear quantization is directly performed on HDR video, then the encoding process is simple, but information about the HDR source is severely damaged

Engineering Contradiction:
Improveencoding simplicityVSAvoidHDR source information
Core Design Contradiction:
Ease of manufactureVSLoss of information

Solution Approach 1:

The patent applies non-linear transfer functions (optical-electro transfer function OETF and electro-optical transfer function EOTF) to transform the HDR video signal parameters before quantization. This changes the luminance distribution characteristics, allowing key luminance segments to be protected while enabling effective quantization to integral bit data for encoding.

Inventive Principle:
Principle #35Parameter changes

2Loss of information

If a transfer function is used for non-linear transfer, then HDR luminance segments are protected, but the conversion process becomes more complex

Engineering Contradiction:
ImproveHDR luminance segment protectionVSAvoidconversion process complexity
Core Design Contradiction:
Loss of informationVSDevice complexity

Solution Approach 1:

The patent performs preliminary non-linear transfer using OETF before quantization and encoding. This preliminary action protects key HDR luminance segments by transforming them into a range suitable for subsequent processing, ensuring that important visual information is preserved before the signal undergoes quantization and compression.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The patent employs a dual transfer function approach with OETF for encoding and EOTF for decoding. The EOTF acts as an inverse transformation that restores the HDR luminance characteristics after quantization, providing a feedback mechanism that maintains image quality throughout the encoding-decoding cycle.

Inventive Principle:
Principle #23Feedback

3Ease of operation

If conventional optical-electro transfer function is used, then the processing is straightforward, but the adaptation to different display devices is limited

Engineering Contradiction:
Improveprocessing simplicityVSAvoiddisplay device adaptation
Core Design Contradiction:
Ease of operationVSAdaptability or versatility

Solution Approach 1:

The patent introduces dynamic parameters into the transfer function, including maximum luminance value (max_display_mastering_luminance) and minimum luminance value (min_display_mastering_luminance) of the display device. These parameters allow the transfer function to dynamically adapt to different display capabilities, enabling the same encoding process to work effectively across diverse display devices with varying luminance ranges.

Inventive Principle:
Principle #15Dynamics

Data Source

PatentUS12284350B2Method and apparatus for processing image signal conversion, and terminal device
Publication Date: 2025.04.22 HUAWEI TECH CO LTD
  • US12284350B2 patent drawing
  • US12284350B2 patent drawing
  • US12284350B2 patent drawing

AI summary

Embodiments of this application provide a method for processing image signal conversion, including: obtaining a primary color component of a to-be-processed pixel; and converting a first value of the primary color component into a second value based on a preset dynamic range conversion model, where the dynamic range conversion model is:L′=F⁡(L)=a×(p×Ln(k1×p-k2)×Ln+k3)m+b,whereL is the first value, L′ is the second value, and k1, k2, k3, a, b, m, n, and p are model parameters.