HEIF Bit-Depth Extension Using Main and Residual Images
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image coding protocols such as HEVC and AV1 have a fixed maximum bit depth per pixel, limiting the representation of high dynamic range images and raw image sensor data, which cannot be fully encoded without loss of precision.
Innovation Solution
A technique that encodes high-bit-depth images by splitting them into a main and extension image, where the main image is encoded at a lower bit depth and the extension image is derived from the residual, allowing reconstruction of the original bit-depth image using legacy-compatible container formats.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If image coding protocols use a fixed maximum bit depth per pixel, then decoder compatibility and simplicity are maintained, but the representation of high dynamic range images and raw image sensor data is limited
Solution Approach 1:
The patent segments the high-bit-depth image into two separate images: a main image encoded at a lower bit depth (compatible with legacy decoders) and an extension image containing the additional bit depth information. This segmentation allows the system to support extended bit depth ranges while maintaining compatibility with existing decoders that only process the main image.
Solution Approach 2:
The extension image is nested within the container format alongside the main image, with the extension image providing supplementary data that enhances the main image's bit depth representation. The container structure allows legacy decoders to ignore the extension image while advanced decoders can utilize it for higher precision reconstruction.
2Measurement precision
If high-bit-depth images are encoded directly, then precision and quality are improved, but compatibility with legacy decoders is lost
Solution Approach 1:
The patent divides the high-precision image data into a main image at legacy-supported bit depth and an extension image containing the additional precision data. This allows legacy decoders to process the main image with guaranteed compatibility while advanced decoders can combine both images to achieve the full precision of the original high-bit-depth source.
Solution Approach 2:
The patent applies different quality levels to different parts of the data structure: the main image provides baseline quality compatible with all decoders, while the extension image provides enhanced quality for advanced decoders. This local quality differentiation ensures broad compatibility while allowing high precision where supported.
3Measurement precision
If a single image is encoded at high bit depth, then precision is maintained, but the container format cannot accommodate legacy-compatible encoding
Solution Approach 1:
The encoding process is segmented into two stages: first encoding the main image at a lower bit depth using well-established legacy codecs, then deriving and encoding the extension image containing the residual high-bit-depth information. This segmentation simplifies the encoding process by allowing the use of mature, well-tested legacy encoding tools while still achieving high precision through the extension image.
Solution Approach 2:
The main image is encoded first using legacy-compatible methods, establishing a foundation that ensures broad compatibility. The extension image is then derived as a supplementary component that adds precision without interfering with the already-encoded main image. This preliminary action approach simplifies the overall encoding workflow.
Data Source
AI summary
Improved image coding techniques for high-bit depth images includes deriving two or more separate lower-bit-depth main and extension images from one high-bit-depth source image, and then encoding them separately as lower-bit-depth images. At a decoder, the separate main and extension images are decoded, and then combined to create a single reconstructed source image. The improved techniques may use lossless or lossy codecs, and may provide some backward compatibility with legacy decoders.