Luminance Block Encoding for HDR Image Compression
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for encoding high-dynamic-range (HDR) images require large compressed sizes, making them unsuitable for applications needing high compression rates, such as transmission contexts, due to the use of traditional encoding schemes that do not efficiently manage the dynamic range of pixel values.
Innovation Solution
A method involving obtaining a luminance residual block by subtracting an average value representation from the luminance component, applying a transfer function to reduce dynamic range, and encoding this block with a traditional encoder, while ensuring perceptual threshold control to minimize visual differences in decoded images.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If traditional encoding schemes are used for HDR images, then encoding simplicity is maintained, but compressed picture size becomes large
Solution Approach 1:
The picture is divided into multiple blocks, and each block is further divided into luminance and chroma components. The luminance block is segmented into residual blocks and average value representations, allowing differential encoding where only changes from the average are stored, reducing overall compression size while maintaining encoding structure.
Solution Approach 2:
The patent applies transfer functions to transform luminance values between different dynamic ranges (LDR and HDR). By changing the parameter representation of luminance values through transfer functions, the encoding adapts to different dynamic range requirements, enabling efficient compression while preserving HDR quality where needed.
2Productivity
If transfer function is applied to reduce dynamic range, then compression efficiency improves, but rendering accuracy in dark and bright ranges deteriorates
Solution Approach 1:
The patent applies different transfer functions to different luminance ranges. Dark luminance values use one transfer function while bright luminance values use another, allowing each range to be optimized for its specific characteristics. This local differentiation maintains signal accuracy in both dark and bright regions while achieving overall compression efficiency.
Solution Approach 2:
The patent dynamically selects and applies appropriate transfer functions based on the luminance range of each block. The encoding process adapts to local luminance characteristics, applying the most suitable transfer function for each region, thereby maintaining precision where needed while achieving compression efficiency overall.
3Measurement precision
If HDR images are encoded with high precision, then visual quality is maintained, but compressed size increases
Solution Approach 1:
The patent applies transfer functions selectively to only the luminance component of the picture, while chroma components are handled separately. This partial application of compression techniques maintains HDR visual quality for the most critical luminance information while achieving compression on the overall image data.
Solution Approach 2:
The patent performs transfer function transformations and block divisions as preliminary steps before the main encoding process. By pre-processing the HDR image into transformed luminance blocks with represented average values, the subsequent encoding operates on already-optimized data, achieving both quality preservation and compression efficiency.
Data Source
Figure 1~2
Figure 3~4
Figure 5~6
AI summary
The present disclosure generally relates to a method an device for encoding a picture block having a luminance component. The method is characterized in that the method comprises: - obtaining (103, 301) a luminance residual block (Lr) by subtracting a first representation (Ll, LlD) of an average value of the luminance component calculated from said luminance component; - obtaining (102) a second representation (f(Ll), f(Ll)D) of the average value by applying a transfer function to the average value (Ll) of the luminance component in order to reduce the dynamic range of the average value; - encoding a luminance block (Le) obtained by adding (105, 302) together the luminance residual block (Lr) and the second representation (f(Ll), f(Ll)D) of the average value. The disclosure further relates to a method and device of decoding a picture block.