OCR Error Correction via Font-Specific Error Models
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current Optical Character Recognition (OCR) systems produce systematic errors due to font characteristics such as kerning, letter weight, and visual contrast, which are critical in business applications requiring high accuracy.
Innovation Solution
A computer-implemented method and system that generates an error model based on the frequency of errors in groundtruth character strings to correct OCR output by modifying characters, using a user-defined pattern and the Wagner-Fischer algorithm for alignment and error prediction.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If OCR systems are used to convert image format text to machine-encoded text, then information entry and processing efficiency is improved, but systematic errors occur due to font characteristics such as kerning, letter weight, and visual contrast
Solution Approach 1:
The patent introduces an intermediary error model that acts as a mediator between the OCR system output and the final text. This error model, built from groundtruth data, corrects systematic errors by predicting likely character confusions based on font characteristics. The intermediary layer transforms the raw OCR output into corrected text without requiring changes to the OCR system itself or manual intervention for each error.
Solution Approach 2:
The patent applies preliminary action by pre-building an error model from groundtruth character strings before actual OCR correction is needed. This error model captures systematic errors associated with specific fonts and characteristics in advance, allowing the system to quickly apply corrections during text conversion without real-time analysis of font characteristics. The Wagner-Fischer algorithm is used to pre-compute alignment and error patterns.
2Measurement precision
If traditional OCR correction methods are used, then some errors are corrected, but systematic errors arising from font characteristics remain uncorrected
Solution Approach 1:
The patent applies parameter changes by incorporating font-specific parameters into the error model. Instead of using a generic correction approach, the system builds error models that capture characteristics specific to different fonts, including kerning patterns, letter weight variations, and visual contrast properties. This allows the correction system to adapt to different font families and styles by selecting the appropriate pre-built error model.
Solution Approach 2:
The patent applies local quality by creating specialized error models for different font characteristics rather than using a single universal correction method. Each error model is tailored to specific font properties and error patterns, allowing the system to apply locally optimized correction strategies for different text sources. The Wagner-Fischer alignment algorithm enables precise local adjustments to character sequences based on the specific error patterns detected.
Data Source
AI summary
A computer implemented method and system for correcting error produced by Optical Character Recognition (OCR) of text contained in an image encoded document. An error model representing frequency and type of errors produced by Optical Character Recognition Engine is generated. An OCR character string generated by OCR is retrieved. A user-defined pattern of a plurality of character strings is retrieved, where each character string represents a possible correct representation of characters in the OCR character string. The OCR character string is compared to each of the above generated character strings and a ‘likelihood score’ is calculated based on the information from the error model. The character string with the highest ‘likelihood score’ is presumed to be the corrected version of the OCR character string.