Document Image Extraction Using Conceptual Feature Groups

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Conventional methods struggle to accurately extract item values from document images, particularly those noted by irregular character strings such as personal names, addresses, or occupations, due to difficulties in evaluating validity using data types.

Innovation Solution

An information processing apparatus that utilizes machine learning to identify and extract candidate character strings by associating them with predefined conceptual feature groups, using a character string classifier and an attribute information table to determine the appropriate attribute groups.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If conventional data type-based extraction methods are used, then regular character strings can be extracted accurately, but irregular character strings cannot be extracted reliably

Engineering Contradiction:

Improveextraction accuracyVSAvoidhandling of irregular character strings

Core Design Contradiction:

Measurement precisionVSAdaptability or versatility

Solution Approach 1:

The patent changes the extraction parameter from data type (conventional method) to conceptual feature groups (invention method). By transforming character strings into feature vectors and comparing them with predefined conceptual feature groups, the system can handle both regular and irregular character strings effectively, resolving the contradiction between extraction accuracy and adaptability

Inventive Principle:

Principle #35Parameter changes

Solution Approach 2:

The patent introduces conceptual feature groups as an intermediary between the extracted character strings and their classification. These feature groups serve as a bridge that enables the system to categorize diverse character strings (both regular and irregular) into appropriate attribute groups, thereby improving both accuracy and versatility

Inventive Principle:

Principle #24Intermediary (Mediator)

2Measurement precision

If manual classification is performed, then accurate categorization can be achieved, but processing time increases significantly

Engineering Contradiction:

Improveclassification accuracyVSAvoidprocessing time

Core Design Contradiction:

Measurement precisionVSLoss of time

Solution Approach 1:

The patent replaces the mechanical manual classification process with an automated computational system. By using machine learning models to compare feature vectors with predefined conceptual feature groups, the system achieves accurate classification automatically, eliminating the need for manual intervention while maintaining high accuracy

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

Solution Approach 2:

The patent performs preliminary action by pre-defining multiple conceptual feature groups and their associated attribute information before the actual extraction process. This preparation enables the system to quickly and accurately classify character strings during processing, reducing both time and effort compared to manual classification

Inventive Principle:

Principle #10Preliminary action

3Productivity

If item values are extracted based on predetermined keywords and data types, then extraction speed is maintained, but extraction accuracy decreases for irregular character strings

Engineering Contradiction:

Improveextraction speedVSAvoidextraction accuracy

Core Design Contradiction:

ProductivityVSMeasurement precision

Solution Approach 1:

The patent changes the extraction parameter from rigid keywords and data types to flexible conceptual feature groups. This allows the system to maintain high extraction speed through automated feature vector comparison while significantly improving accuracy for irregular character strings that cannot be captured by predetermined patterns

Inventive Principle:

Principle #35Parameter changes

Data Source

PatentUS12481831B2Information processing apparatus, information processing method, storage medium, and learning apparatus

Publication Date: 2025.11.25 CANON KK

AI summary

Provided is an information processing apparatus that extracts a character string to be a candidate of an item value corresponding to a predetermined item, from among multiple character strings included in a document image to be processed. The information processing apparatus first obtains information indicating a conceptual feature group of a preset extraction target, then identifies a conceptual feature group to which each of the character strings included in the document image to be processed belongs from among predetermined multiple conceptual feature groups based on a feature vector corresponding to each character string, and then extracts the character string whose identified conceptual feature group is the same as the conceptual feature group associated with the same attribute information as the conceptual feature group of the extraction target, from among the character strings included in the document image, as the candidate.