HDR to SDR Tone Mapping via Cascade Networks
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing technologies for converting HDR images to SDR images often result in visible tone mapping artifacts, particularly in high-chroma gradation scenes, leading to degradation of image quality.
Innovation Solution
An image processing apparatus and method that compresses HDR image data to a narrower dynamic range while maintaining a predetermined tone curve, using PQ scaling processing to prioritize natural gradation and potentially switching to linear scaling for high luminance areas, thereby minimizing color deviation and image quality degradation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If switching between RGB tone mapping processing and YUV tone mapping processing is performed for HDR to SDR conversion, then the conversion can be completed across different luminance ranges, but visible switching artifacts appear in high-chroma gradation scenes leading to image quality degradation
Solution Approach 1:
The patent divides the dynamic range compression into multiple stages using a cascade structure with multiple tone mapping networks. Each network processes a specific portion of the luminance range, avoiding the need to switch between different processing modes (RGB vs YUV) that causes visible artifacts. This segmentation allows continuous processing across the full luminance spectrum while maintaining image quality in high-chroma gradation scenes.
2Device complexity
If a single tone mapping method is used for the entire luminance range, then processing is simple and continuous, but it cannot optimally handle both dark and high luminance portions simultaneously
Solution Approach 1:
The patent employs a dynamic cascade structure where multiple tone mapping networks are activated based on the input luminance characteristics. The system dynamically selects and combines the outputs of different networks to optimally handle various luminance portions. This dynamic approach allows the system to adapt to different scene conditions while maintaining continuous processing without abrupt switches, thereby preserving gradation quality.
3Adaptability or versatility
If HDR image data is compressed to SDR dynamic range, then the image can be displayed on SDR monitors, but color deviation occurs and the hue of subjects varies
Solution Approach 1:
The patent utilizes multiple tone mapping networks with different characteristic parameters optimized for specific luminance ranges. By adjusting and combining the outputs of these networks based on input characteristics, the system maintains color accuracy throughout the compression process. The parameter-based approach allows flexible adaptation to preserve hue information while achieving SDR compatibility for display on various monitors.
Data Source
AI summary
There is provided an image processing apparatus. An obtaining unit obtains image data having a first dynamic range, the image data having been applied with a predetermined tone curve, the predetermined tone curve being a tone curve representing visual characteristics based on a relationship between the first dynamic range and an absolute luminance that is output. A first compression unit compresses the image data from the first dynamic range to a second dynamic range narrower than the first dynamic range while maintaining the predetermined tone curve.