String Correction via Minimum Cost Edit Operations
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current Optical Character Recognition (OCR) systems face challenges in correcting errors in non-word strings, such as vehicle license plate numbers and serial numbers, due to structural similarities and non-ideal image acquisition conditions, where traditional dictionary lookup methods are inadequate.
Innovation Solution
A method is proposed to determine a set of edit operations (deletion, insertion, substitution) to transform a string into a string template by calculating the minimum cost of edit operations using dynamic programming, allowing for the correction of strings to satisfy specific templates like vehicle license plate formats.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional dictionary lookup methods are used for string correction, then correction is effective for word-type strings, but it is inadequate for non-word strings such as license plate numbers and serial numbers
Solution Approach 1:
The invention changes the fundamental parameter of correction methodology from dictionary-based word matching to template-based structural transformation. By defining templates that represent valid formats (e.g., license plate patterns, serial number structures) and applying edit operations to transform erroneous strings into valid templates, the system achieves versatility across both word-type and non-word-type strings while maintaining correction reliability
Solution Approach 2:
The template-based correction system creates a universal framework that can handle multiple types of strings (words, license plates, serial numbers, IDs) through a single unified approach. The system uses general edit operations (insertion, deletion, substitution) combined with template matching to correct any string type, eliminating the need for separate correction mechanisms for different string categories
2Productivity
If OCR is performed under non-ideal conditions, then image acquisition is faster and more practical, but OCR errors increase due to lighting variations, occlusion, and character degradation
Solution Approach 1:
The invention applies preliminary correction actions by defining valid templates beforehand and using them to guide the correction process. Before finalizing the OCR result, the system performs template-based validation and applies minimal edit operations to transform erroneous strings into valid ones, thereby compensating for recognition errors caused by non-ideal imaging conditions while maintaining fast acquisition speeds
Solution Approach 2:
The system implements feedback by comparing the OCR-generated string against predefined templates and using the discrepancy information to guide correction. The template matching process provides feedback on which characters or positions are erroneous, allowing the system to apply targeted edit operations that restore the string to a valid format, thus improving accuracy without sacrificing productivity
Data Source
AI summary
Determining a set of edit operations to perform on a string, such as one generated by optical character recognition, to satisfy a string template by determining a minimum cost of performing edit operations on the string to satisfy the string template and then determining the set of edit operations corresponding to the minimum cost. Transforming a string to satisfy one or more string templates by determining a minimum cost of performing edit operations on the string to satisfy one or more string templates, selecting one or more minimum costs, determining a set of edit operations corresponding to the minimum costs, and then performing the set of edit operations on the string. Determining a minimum cost of performing edit operations on a string to satisfy a string template by determining set costs of performing sets of edit operations using costs associated with edit operations of the set and determining the minimum cost using the set costs.