Grid-Based Template Optical Character Recognition
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Traditional optical character recognition (OCR) systems sacrifice speed to ensure accuracy, particularly when distinguishing between similar characters like '0' and 'O', leading to time-consuming multi-step template-based matching procedures.
Innovation Solution
The use of a grid-based template that differentiates characters based on centerline height differences, employing a software algorithm to quickly identify numerals and letters by limiting processing to specific grid sections and coordinate locations, and utilizing a lookup table for rapid recognition.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional template-based pattern matching algorithm is used to ensure accurate character recognition, then recognition accuracy is improved, but recognition speed deteriorates
Solution Approach 1:
The patent divides the character recognition process into two distinct phases: a fast pre-processing phase that performs initial character identification using simplified criteria, and a secondary verification phase that applies more rigorous pattern matching only when needed. This segmentation allows the system to achieve high speed for most characters while maintaining accuracy for ambiguous cases.
Solution Approach 2:
The patent applies partial action by performing only the necessary level of analysis for each character. For clearly distinguishable characters, the system performs minimal processing (partial action) to achieve fast recognition. For ambiguous characters requiring higher accuracy, the system performs more comprehensive analysis. This avoids the excessive action of applying full pattern matching to all characters uniformly.
2Measurement precision
If multi-step template matching procedure is used to distinguish similar characters like '0' and 'O', then recognition accuracy is improved, but processing time increases
Solution Approach 1:
The patent performs preliminary characterization of all characters using a fast algorithm that identifies key distinguishing features such as centerline height, aspect ratio, and pixel density. This preliminary action creates a preliminary classification that allows the system to quickly differentiate between similar characters like '0' and 'O' based on their geometric properties before applying more time-consuming pattern matching, thus reducing overall processing time while maintaining accuracy.
Solution Approach 2:
The patent transforms the character recognition problem from pure pattern matching to a parameter-based differentiation approach. By extracting and comparing specific geometric parameters (centerline height, width-to-height ratio, pixel distribution patterns), the system can rapidly distinguish between similar characters. This parameter-based approach changes the recognition criteria from holistic pattern matching to specific measurable attributes, enabling faster differentiation of ambiguous characters.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
The present disclosure is generally directed to systems and methods for executing optical character recognition faster than at least some traditional OCR systems, without sacrificing recognition accuracy. Towards this end, various exemplary embodiments involve the use of a bounding box and a grid-based template to identify certain unique aspects of each of various characters and/or numerals. For example, in one embodiment, the grid-based template can be used to recognize a numeral and/or a character based on a difference in centerline height between the numeral and the character when a monospaced font is used. In another exemplary embodiment, the grid-based template can be used to recognize an individual digit among a plurality of digits based on certain parts of the individual digit being uniquely located in specific portions of the grid-based template.