Character Recognition Device Using Dual Scoring for Overlap Prevention
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing character recognition techniques struggle to accurately recognize characters in input images without incorrectly overlapping characters or skipping over them.
Innovation Solution
A character recognition device comprising a first score calculation unit, a character region estimation unit, a second score calculation unit, and a selection unit, which calculates scores to assess the likelihood and consistency of character strings within an input image, thereby selecting the most accurate character strings.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If characters are recognized without explicitly dividing boundaries between characters, then recognition accuracy is improved, but characters may be recognized overlappingly or skipped
Solution Approach 1:
The patent segments the character recognition process into multiple independent modules: a character string candidate generation unit that produces multiple candidate character strings, a character region estimation unit that estimates regions for each character, and a selection unit that selects the correct character string. This segmentation allows the system to evaluate multiple hypotheses simultaneously and choose the most accurate one, resolving the contradiction between recognizing characters without boundary division and avoiding overlapping/skipping errors.
Solution Approach 2:
The patent implements feedback through the selection unit that uses both the first score (likelihood of character string) and the second score (consistency of character regions) to feedback-select the correct character string from candidates. This feedback mechanism allows the system to correct potential overlapping or skipping errors by comparing multiple candidates and selecting the one that best fits both likelihood and spatial consistency criteria.
2Reliability
If multiple candidate character strings are generated and evaluated, then recognition correctness is improved, but computational complexity increases
Solution Approach 1:
The patent changes parameters by introducing two scoring metrics: the first score for likelihood of character string and the second score for consistency of character regions. By transforming the selection criterion into a multi-parameter evaluation system, the system can efficiently filter and rank candidate character strings without exhaustive computation, reducing computational complexity while maintaining high recognition correctness.
Data Source
AI summary
A character recognition device of an embodiment includes a first score calculation unit, a character region estimation unit, a second score calculation unit, and a selection unit. The first score calculation unit calculates a first score indicating a likelihood of a character string, or the first score for each of a plurality of candidate character strings which are candidates for character strings included in an input image. The character region estimation unit estimates a region corresponding to each character included in the candidate character string among regions of the input image. The second score calculation unit calculates a second score indicating a consistency of characters included in the candidate character string on the basis of the estimated region. The selection unit selects one or more character strings from among the plurality of candidate character strings on the basis of the calculated first score and the calculated second score.