Bit-Exact Lossless Image Coding Residual System

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

JPEG-2000 lacks a method to combine lossy and lossless compression techniques effectively, limiting its compression efficiency and dynamic range capabilities, especially in handling floating-point values and extended brightness ranges.

Innovation Solution

A lossy compression process is used to generate a compressed image that can be losslessly reconstructed by creating a lossless residual layer using a sorted list of pixel values and bin indices, allowing for bit-exact reproduction of the original image, regardless of integer or floating-point pixel values.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If lossless coding methods (JPEG-2000 DWT 5/3) are used, then bit-exact reproduction is achieved, but compression efficiency is limited to about 2.3:1

Engineering Contradiction:

Improvebit-exact reproductionVSAvoidcompression efficiency

Core Design Contradiction:

ReliabilityVSProductivity

Solution Approach 1:

The invention segments the image coding process into two independent parts: a lossy JPEG-2000 base layer for compression and a lossless residual layer for bit-exact reproduction. The residual layer encodes only the difference between the original and lossy-decoded images, allowing both high compression ratio and perfect reconstruction to coexist.

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The invention merges lossy and lossless coding techniques by combining the JPEG-2000 lossy compression with a custom lossless residual encoding. The residual layer uses a sorted list of pixel values and bin indices to achieve lossless coding of the difference data, creating a hybrid system that achieves both compression efficiency and bit-exact reproduction.

Inventive Principle:

Principle #5Merging (Combining)

2Ease of manufacture

If integer pixel values are used, then coding simplicity is maintained, but dynamic range is limited and precision is restricted

Engineering Contradiction:

Improvecoding simplicityVSAvoiddynamic range and precision

Core Design Contradiction:

Ease of manufactureVSMeasurement precision

Solution Approach 1:

The invention changes the data type parameter from fixed-precision integers to floating-point values for the residual encoding. This allows the system to handle extended dynamic range and high precision pixel values while maintaining coding simplicity through the use of a sorted list and bin index approach. The floating-point residual values are then quantized to the appropriate precision for storage and transmission.

Inventive Principle:

Principle #35Parameter changes

3Productivity

If lossy compression is applied, then compression ratio is improved, but bit-exact reproduction is lost

Engineering Contradiction:

Improvecompression ratioVSAvoidbit-exact reproduction

Core Design Contradiction:

ProductivityVSReliability

Solution Approach 1:

The invention performs a preliminary lossy compression using JPEG-2000 to achieve high compression ratio, then uses the original image and the lossy-decoded image to compute a residual. This residual contains only the lossy artifacts and is encoded losslessly, allowing perfect reconstruction when combined with the lossy base layer.

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The invention introduces a residual layer as an intermediary between the lossy compression and the original image. This residual layer captures the differences and errors introduced by lossy compression, allowing the system to maintain bit-exact reproduction by adding the residual back to the lossy-decoded image.

Inventive Principle:

Principle #24Intermediary (Mediator)

4Measurement precision

If regional integer scale factors are used, then integer dynamic range is extended, but precision is clipped and negative values cannot be represented

Engineering Contradiction:

Improveinteger dynamic rangeVSAvoidprecision clipping and negative value limitation

Core Design Contradiction:

Measurement precisionVSLoss of information

Solution Approach 1:

The invention replaces the integer-based scaling mechanism with a floating-point-based residual encoding system. This substitution eliminates the limitations of integer arithmetic, allowing the system to represent negative values, maintain full precision without clipping, and handle extended dynamic range naturally through floating-point arithmetic.

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

Data Source

PatentUS7769239B2Efficient bit-exact lossless image coding residual system

Publication Date: 2010.08.03 DEMOS GARY

AI summary

A system and method for bit-exact lossless compression coding of still and moving images. An original image is encoded to produce a lossy compressed image, and is also used to construct a sorted list of bin values to which the lossy image is compared to generate output from which the original image can be losslessly reconstructed. The similarity of corresponding pixel values between the lossy image and the original image permits efficient generation of a lossless residual that allows for bit-exact reproduction of the original image. Pixel values nearly equidistant from two sorted list values are on a “cusp” and either directly coded or flagged as cusp values from which corresponding lossless values can be re-determined. Non-cusp pixel values are coded as the difference between the index of the sorted bin having a value nearest to the floating point value being processed, and the index of the sorted bin containing the exact pixel value of the original image.