Iris Code Segmentation for Database Search Speed

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current iris recognition systems face inefficiencies in matching iris templates within large databases, as the process is time-consuming due to the analysis of raw iris images, requiring significant computational resources and often involving unnecessary comparisons.

Innovation Solution

The system employs a multi-stage clustering approach using phase, radial, and angular information to group iris code bits into subclasses, allowing for progressive matching that reduces the number of templates to be compared, thereby compressing the feature vector and expediting the search process.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If traditional iris template matching is performed on raw iris images, then comprehensive feature analysis is achieved, but the matching process becomes time-consuming and computationally intensive

Engineering Contradiction:
Improvefeature analysis completenessVSAvoidmatching process time
Core Design Contradiction:
Measurement precisionVSLoss of time

Solution Approach 1:

The patent segments the iris code into multiple bit sub-classes based on angular positions (e.g., 0-45°, 45-90°, etc.). This segmentation allows the system to perform progressive matching by comparing subsets of bits rather than the entire iris code at once, significantly reducing computational complexity while maintaining matching accuracy.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent performs preliminary clustering of iris codes into angular-based groups before the actual matching process. By pre-organizing the database templates into angular sub-classes, the system can quickly narrow down potential matches without performing comprehensive comparisons on all templates, thus reducing matching time while preserving feature analysis completeness.

Inventive Principle:
Principle #10Preliminary action

2Reliability

If comprehensive iris template matching is performed across the entire database, then accurate identification is achieved, but the computational burden increases significantly

Engineering Contradiction:
Improveidentification accuracyVSAvoidcomputational efficiency
Core Design Contradiction:
ReliabilityVSProductivity

Solution Approach 1:

The patent divides the database of iris templates into multiple angular sub-classes (e.g., based on 45° angular segments). During matching, the query iris code is also segmented and compared only against relevant sub-classes rather than the entire database. This segmentation maintains identification accuracy by preserving all necessary feature comparisons while dramatically improving computational efficiency by reducing the search space.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent performs preliminary clustering of templates into angular groups before matching. This pre-processing step organizes the database structure so that during identification, the system only needs to search within specific angular sub-classes rather than comparing against all templates. This preliminary organization maintains reliability by ensuring all relevant comparisons are made while boosting productivity by reducing the number of comparisons required.

Inventive Principle:
Principle #10Preliminary action

3Measurement precision

If all iris code bits are used for matching, then complete feature comparison is achieved, but the number of operations increases from 0.091 to 2.4 Mega bytes

Engineering Contradiction:
Improvefeature comparison completenessVSAvoidcomputational operations
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent segments the iris code bits into angular sub-classes (e.g., dividing the 360° iris into multiple angular segments). Instead of comparing all bits across the entire iris code, the system performs matching on segmented subsets. This approach maintains feature comparison completeness by preserving all angular information while reducing computational operations by processing smaller bit subsets in parallel or sequentially.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent performs preliminary clustering of iris codes into angular groups based on dominant angular positions. This pre-processing step identifies which angular sub-classes are most relevant for each iris code, allowing the matching process to focus only on those specific bit subsets rather than processing the entire iris code. This preliminary action maintains measurement precision by preserving all necessary feature information while reducing device complexity by eliminating redundant computational operations.

Inventive Principle:
Principle #10Preliminary action

Data Source

PatentEP1991947B1Indexing and database search system
Publication Date: 2020.04.29 GENTEX CORP
  • EP1991947B1 patent drawingFigure 1a
  • EP1991947B1 patent drawingFigure 1b
  • EP1991947B1 patent drawingFigure 2

AI summary

An indexing and database search system having a system to expedite a process of matching a subject to millions (more or less) of templates within a database. Such matching or retrieval may use a subject code as a query or probe and then find a similarity measure for the features of codes or templates in the database. A multi-stage clustering process may be used to compress codes and/or templates.