Ridge Image Selection for Machine Learning Using Clear Zone Scoring

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing techniques for processing information related to ridges, such as fingerprints or palm prints, face challenges in accurately extracting feature quantities and distinguishing clear and unclear zones within ridge images, which affects the effectiveness of machine learning models.

Innovation Solution

An information processing system that includes a storage unit for storing area information about clear and unclear zones in ridge images, an extraction unit to identify these zones, a calculation unit to score the images based on this information, and a selection unit to choose images suitable for machine learning, thereby improving the accuracy of feature extraction and model learning.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Quantity of substance

If machine learning is performed using all available ridge images, then the quantity of training data increases, but the accuracy of feature extraction decreases due to inclusion of unclear zone images

Engineering Contradiction:

Improvequantity of training dataVSAvoidaccuracy of feature extraction

Core Design Contradiction:

Quantity of substanceVSMeasurement precision

Solution Approach 1:

The patent segments the training data by dividing ridge images into clear zones and unclear zones using area information. The extraction unit processes only clear zone images for machine learning, separating useful training data from problematic data that would degrade model performance.

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The patent applies local quality by evaluating each ridge image's clear zone area ratio and using this local characteristic to determine suitability for training. Images with higher clear zone proportions are selected for machine learning, ensuring that the training data locally satisfies the quality requirement for accurate feature extraction.

Inventive Principle:

Principle #3Local quality

2Measurement precision

If clear zone and unclear zone are distinguished using complex image processing, then the accuracy of zone identification improves, but the device complexity increases

Engineering Contradiction:

Improveaccuracy of zone identificationVSAvoidcomplexity of processing system

Core Design Contradiction:

Measurement precisionVSDevice complexity

Solution Approach 1:

The patent applies preliminary action by pre-storing area information indicating clear zones and unclear zones in a storage unit before machine learning. This pre-processing step provides the extraction unit with ready-to-use zone definitions, eliminating the need for complex real-time image processing during training and reducing system complexity.

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The patent uses area information as an intermediary between the ridge images and the machine learning process. This intermediary data structure stores pre-analyzed zone information, allowing the extraction unit to identify clear and unclear zones without performing complex image processing operations directly on the ridge images.

Inventive Principle:

Principle #24Intermediary (Mediator)

3Quantity of substance

If ridge images with unclear zones are used for training, then the quantity of training data increases, but the reliability of the machine learning model decreases

Engineering Contradiction:

Improvequantity of training dataVSAvoidreliability of machine learning model

Core Design Contradiction:

Quantity of substanceVSReliability

Solution Approach 1:

The patent applies partial action by selectively using only the clear portion (clear zones) of ridge images for machine learning training. Instead of using entire images including unclear portions, the system extracts and trains only on the reliable clear zone portions, ensuring high reliability while maintaining sufficient training data quantity.

Inventive Principle:

Principle #16Partial or excessive action

Solution Approach 2:

The patent implements feedback by using the area information to evaluate each ridge image's suitability for training. The extraction unit receives feedback from the area information about clear and unclear zone distributions, and this feedback determines which images are selected for machine learning, ensuring that only reliable images are used.

Inventive Principle:

Principle #23Feedback

Data Source

PatentUS12333848B2Information processing system, information processing apparatus, information processing method, and recording medium

Publication Date: 2025.06.17 NEC CORP

AI summary

An information processing system includes: a storage unit that stores first area information indicating a clear zone in which a feature quantity of a ridge in a ridge image can be extracted, and second area information indicating an unclear zone that is an area other than the clear zone, in association with each of a plurality of ridge images; an extraction unit that extracts third area information indicating the clear zone, and fourth area information indicating the unclear zone, from the ridge image; a calculation unit that calculates a first score based on the first area information and the second area information, and a second score based on the third area information and the fourth area information; and a selection unit that selects the ridge image to be used for machine learning of the extraction unit, on the basis of the first score and the second score.