Word Recognition Apparatus Unmatched Character Evaluation

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing word recognition systems face challenges in accurately distinguishing similar words due to insufficient recognition accuracy when a plurality of words with matched portions are selected as candidates for recognition, as the degree of difference between similar words is not effectively calculated.

Innovation Solution

A word recognition apparatus and method that focuses on evaluating the difference in recognition results by calculating evaluation values for unmatched character portions between word candidates, allowing for more precise differentiation and improved accuracy by recalculating evaluation values for unmatched portions.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Device complexity

If the degree of difference between similar words is calculated using the method in Patent Literature 1, then the word recognition process can be simplified, but the recognition accuracy deteriorates because similar words with matched portions cannot be effectively distinguished

Engineering Contradiction:
Improveword recognition process complexityVSAvoidrecognition accuracy
Core Design Contradiction:
Device complexityVSMeasurement precision

Solution Approach 1:

The patent segments the word comparison process into two distinct stages: first comparing matched character strings to identify common portions, then comparing unmatched character strings to differentiate similar words. This segmentation allows the system to simplify the overall process while maintaining high accuracy by focusing computational effort only where needed.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent applies local quality by differentiating the evaluation approach for matched versus unmatched portions. Matched portions are evaluated for commonality, while unmatched portions are evaluated for差异性 using character image similarity. This localized differentiation enables accurate distinction of similar words without requiring complex processing of all word components.

Inventive Principle:
Principle #3Local quality

2Measurement precision

If all character strings are compared between word candidates, then comprehensive comparison is achieved, but computational efficiency deteriorates due to redundant comparisons of matched portions

Engineering Contradiction:
Improvecomparison completenessVSAvoidcomputational efficiency
Core Design Contradiction:
Measurement precisionVSProductivity

Solution Approach 1:

The patent extracts and removes matched character strings from the comparison set, retaining only unmatched portions for differential evaluation. This extraction eliminates redundant comparisons of identical character sequences while preserving the essential information needed to distinguish similar words, thereby significantly improving computational efficiency.

Inventive Principle:
Principle #2Taking out (Extraction)

Solution Approach 2:

The patent applies partial action by performing detailed comparison only on the necessary unmatched portions rather than all character strings. This selective approach maintains sufficient comparison completeness for accurate word differentiation while reducing unnecessary computational overhead.

Inventive Principle:
Principle #16Partial or excessive action

Data Source

PatentEP2341467B1Word recognition device, method, non-transitory computer readable medium storing program and shipped item classification device
Publication Date: 2019.12.18 NEC CORP
  • EP2341467B1 patent drawingFigure 1~2
  • EP2341467B1 patent drawingFigure 3
  • EP2341467B1 patent drawingFigure 4

AI summary

To improve the recognition accuracy even when a plurality of words including a matched portion are selected as candidates for recognition. A word recognition apparatus according to the present invention includes input means for inputting a word image representing a plurality of characters; word candidate selection means for recognizing the word image input by the input means and selecting a first word candidate and a second word candidate based on a plurality of words registered in a word dictionary; and verification means for comparing the first word candidate and the second word candidate character by character and verifying a likelihood of the first word candidate based on an evaluation value obtained when the word image is recognized by characters determined as unmatched.