Compression Coding Without Identification Bits for Better Ratios

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Conventional compression techniques, such as LZ77 and Huffman coding, suffer from a decrease in compression rate due to the inclusion of identification bits in compression codes, particularly when dealing with data having high appearance frequencies.

Innovation Solution

A compression program that searches for the longest matching character strings and generates frequency information to associate compression codes with either the data or position information based on the length of the matching strings, eliminating the need for identification bits in the compression codes.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If identification bits are included in compression codes to distinguish between different encoding types, then the decoding process can correctly interpret the compression data, but the compression rate decreases due to the additional bits

Engineering Contradiction:
Improvedecoding accuracyVSAvoidcompression rate
Core Design Contradiction:
ReliabilityVSProductivity

Solution Approach 1:

The patent extracts and removes the identification bit from the compression code structure. Instead of using a separate identification bit to distinguish between literal bytes and match references, the system uses the Huffman code structure itself to encode this information implicitly through the frequency tables and code assignment patterns, thereby eliminating the need for explicit identification bits and improving compression rate.

Inventive Principle:
Principle #2Taking out (Extraction)

Solution Approach 2:

The patent merges the identification function into the existing Huffman coding structure. By combining the length encoding and type identification into a unified Huffman code system with separate frequency tables for different code types, the system achieves both compression and reliable decoding without requiring additional identification bits.

Inventive Principle:
Principle #5Merging (Combining)

2Productivity

If separate frequency tables are used for different code types (literal bytes and match references), then the compression can be optimized for each type, but the device complexity increases

Engineering Contradiction:
Improvecompression optimizationVSAvoidfrequency table management
Core Design Contradiction:
ProductivityVSDevice complexity

Solution Approach 1:

The patent segments the frequency table management into two distinct tables: one for literal bytes and one for match references. This segmentation allows independent optimization and frequency counting for each code type, enabling better compression ratios while maintaining clear separation of concerns in the implementation.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent creates a universal Huffman decoding framework that can handle multiple code types through a unified process. The same decoding logic is applied to both literal byte codes and match reference codes, with the only difference being the frequency table used for probability calculation. This multi-functional approach reduces implementation complexity despite using separate frequency tables.

Inventive Principle:
Principle #6Universality (Multi-functionality)

Data Source

PatentUS9479195B2Non-transitory computer-readable recording medium, compression method, decompression method, compression device, and decompression device
Publication Date: 2016.10.25 FUJITSU LTD
  • US9479195B2 patent drawing
  • US9479195B2 patent drawing
  • US9479195B2 patent drawing

AI summary

At a preliminary stage, a compressing unit generates frequency information, outputs a compression code associated with a piece of first data of the longest matching character string among the pieces of first data contained in the frequency information, when the longest matching character string has a length smaller than the predetermined length and outputs a compression code associated with a piece of position information matching with position information about the longest matching character string among the pieces of position information about the second data contained in the frequency information and a compression code associated with length information about the longest matching character string among the pieces of first data contained in the frequency information, when the longest matching character string has a length equal to or larger than the predetermined length.