Sign Bit Removal in Digital Data Compression

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing image compression methods face challenges in achieving lossless compression due to high complexity and resource requirements, especially when dealing with high entropy images, and lack efficient methods for pre-calculating data budgets and handling sign bits effectively.

Innovation Solution

A method that groups coefficients, determines the Greatest Coded Line Index (GCLI), applies quantization levels, and separates sign bits from magnitude bits to create a compressed data set with a meta-data output, allowing for efficient data budget calculation and reduced complexity, enabling pre-calculation of the data budget and simplifying hardware implementation.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If all signs of coefficients are packed in the output data set, then the sign information is preserved, but the data budget increases and compression efficiency decreases

Engineering Contradiction:
Improvesign information preservationVSAvoiddata budget
Core Design Contradiction:
ReliabilityVSQuantity of substance

Solution Approach 1:

The patent extracts only the necessary sign bits from the coefficient data. Instead of packing all signs, it identifies and removes redundant sign bits for zero coefficients, keeping only the essential sign information needed for reconstruction. This selective extraction reduces the data budget while preserving necessary sign information.

Inventive Principle:
Principle #2Taking out (Extraction)

Solution Approach 2:

The patent discards redundant sign bits that can be inferred or are unnecessary for reconstruction, particularly for zero coefficients. The essential sign information is recovered and preserved in the compressed data, allowing accurate reconstruction without transmitting all original sign bits.

Inventive Principle:
Principle #34Discarding and recovering

2Productivity

If arithmetic encoding is used to process all passes on codeblocks, then compression is achieved, but the complexity and number of operations increase significantly

Engineering Contradiction:
Improvecompression capabilityVSAvoidarithmetic encoder complexity
Core Design Contradiction:
ProductivityVSDevice complexity

Solution Approach 1:

The patent extracts and processes only the essential magnitude and sign bit information, separating it from the full arithmetic encoding process. By focusing on these key elements and using simpler encoding methods for them, the overall system complexity is reduced while maintaining compression effectiveness.

Inventive Principle:
Principle #2Taking out (Extraction)

Solution Approach 2:

The patent employs simpler, less complex encoding approaches for the magnitude and sign bits rather than using computationally intensive arithmetic encoding for all data. This substitutes expensive operations with cheaper, more efficient ones for specific data portions.

Inventive Principle:
Principle #27Cheap short-living objects (Disposable)

3Productivity

If many entropy encoders are run in parallel to achieve typical video resolutions and framerates, then the processing speed is sufficient, but the logic gate count, buffering and power consumption increase hugely

Engineering Contradiction:
Improveprocessing speedVSAvoidlogic gate count and power consumption
Core Design Contradiction:
ProductivityVSDevice complexity

Solution Approach 1:

The patent segments the coefficient data into magnitude parts and sign bit parts, processing them separately with optimized methods. This segmentation allows each part to be handled more efficiently, reducing the overall computational burden and hardware requirements compared to processing all data uniformly through complex entropy encoders.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent extracts and separately handles the sign bits, removing them from the main compression pipeline that would otherwise require complex parallel entropy encoding. This extraction simplifies the hardware architecture by eliminating the need for numerous complex entropy encoder units.

Inventive Principle:
Principle #2Taking out (Extraction)

4Quantity of substance

If quantization is applied to reduce information, then the bitrate is controlled, but the image quality deteriorates

Engineering Contradiction:
Improvebitrate controlVSAvoidimage quality
Core Design Contradiction:
Quantity of substanceVSManufacturing precision

Solution Approach 1:

The patent extracts and preserves the sign bit information separately, ensuring that this critical information is not lost during quantization. By maintaining sign information alongside quantized magnitude data, the reconstruction accuracy is improved compared to standard quantization alone.

Inventive Principle:
Principle #2Taking out (Extraction)

Solution Approach 2:

The patent changes the representation parameters by separating magnitude and sign bits, and by using GCLI to dynamically adjust the number of magnitude bits. This parameter adaptation allows optimized bit allocation that maintains quality while controlling bitrate.

Inventive Principle:
Principle #35Parameter changes

Data Source

PatentEP3343921B1Method and device for digital data compression with sign bit removal
Publication Date: 2022.04.13 INTOPIX
  • EP3343921B1 patent drawingFigure 1
  • EP3343921B1 patent drawingFigure 2
  • EP3343921B1 patent drawingFigure 3

AI summary

The invention provides a method and device for compressing a sequence of signed coefficients wherein coefficients are grouped, for each group, the greatest coded line index (GCLI) is determined, a quantization level t is applied and only the GCLI to t+1 weight bits of the coefficients are copied into the output stream together with the value of the GCLI. According to the invention, sign bits are not coded for coefficient having a zero quantized value. The invention also provides a method for determining a data budget required for compressing a sequence of signed coefficients, in dependence of a quantization level, and a method for compressing an image within the limits of a given data budget constraint. The invention provides a compressed data set coding an image in an efficient way.