Image Bit-Depth Mapping Using Sub-Ranges for Hardware-Limited Coding

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing image processing technologies face challenges in efficiently encoding and reconstructing images with high bit depth due to the lack of hardware support, leading to suboptimal trade-offs between coding complexity, rate-distortion relation, and processing capabilities.

Innovation Solution

The method involves dividing the image into sub-ranges with lengths smaller than the original range, associating sample values with these sub-ranges, and generating a transformed representation for each sample, which is then encoded and decoded using association information.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If high bit depth images are encoded directly, then measurement precision is maintained, but device complexity increases due to lack of hardware support

Engineering Contradiction:
Improvebit depthVSAvoidhardware support
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent divides the high bit depth range into multiple sub-ranges, each mapping to a lower bit depth representation. This segmentation allows the image data to be processed in manageable chunks that are compatible with existing hardware, while still preserving the overall precision of the original high bit depth image through the sub-range mapping mechanism.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent transforms the bit depth parameter by mapping sample values from a first bit depth to a second bit depth using sub-range associations. This parameter change enables the image data to be represented in a format that is more compatible with hardware chipsets while maintaining the essential information through the sub-range mapping.

Inventive Principle:
Principle #35Parameter changes

2Device complexity

If high bit depth images are encoded with lower bit depth, then device complexity is reduced, but measurement precision is lost

Engineering Contradiction:
Improvehardware supportVSAvoidbit depth
Core Design Contradiction:
Device complexityVSMeasurement precision

Solution Approach 1:

The patent introduces sub-range association information as an intermediary between the original high bit depth image and the lower bit depth representation. This intermediary contains mapping data that enables accurate reconstruction of the original image, preventing loss of measurement precision while allowing the use of lower bit depth for encoding.

Inventive Principle:
Principle #24Intermediary (Mediator)

Solution Approach 2:

The patent performs preliminary sub-range mapping and generates association information before the actual encoding process. This preliminary action prepares the data in a way that preserves precision information, which is then used during decoding to accurately reconstruct the original high bit depth image.

Inventive Principle:
Principle #10Preliminary action

3Loss of information

If sub-range transformation is applied, then rate-distortion performance is improved, but coding complexity increases

Engineering Contradiction:
Improverate-distortionVSAvoidcoding complexity
Core Design Contradiction:
Loss of informationVSDevice complexity

Solution Approach 1:

The patent extracts the sub-range association information from the image data and stores it separately. This extraction allows the main image encoding to use simpler lower bit depth representations while the extracted association information is used during decoding to restore the original precision, thereby improving rate-distortion performance without excessively increasing coding complexity.

Inventive Principle:
Principle #2Taking out (Extraction)

Data Source

PatentEP4704419A1Method and apparatus for processing an image
Publication Date: 2026.03.04 BEIJING XIAOMI MOBILE SOFTWARE CO LTD
  • EP4704419A1 patent drawingFigure 1
  • EP4704419A1 patent drawingFigure 2
  • EP4704419A1 patent drawingFigure 3

AI summary

A method of processing comprising a plurality of samples having sample values associated therewith, wherein the sample values have a first bit depth representing a first range, comprises: obtaining a plurality of sub-ranges of the first range, wherein lengths of the sub-ranges are smaller than a length of the first range; generating association information, which associates each of the sample values with one of the sub-ranges; and determining a transformed representation of the image based on the sample values and based on the sub-range associated with each of the sample values, and wherein the transformed representation comprises a transformed sample value for one of the samples of the image.