Form Recognition via Line Segment Pair Filtering
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In form recognition methods, when the number of line segments associated with a single form ID is large, the processing time for detailed identification becomes lengthy due to the inclusion of non-corresponding line segment pairs, reducing efficiency.
Innovation Solution
A form recognition method that involves extracting a candidate form ID by correlating line segments using a coarse classification dictionary, limiting the number of corresponding line segment pairs to reduce processing load and enhance efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If detailed identification processing is performed by all form IDs and the form to be recognized, then recognition accuracy is improved, but processing time increases significantly
Solution Approach 1:
The patent divides the form recognition process into two distinct stages: coarse classification and detailed identification. In the coarse classification stage, the form is classified into a candidate group based on overall layout features. In the detailed identification stage, only the candidate forms from this reduced group undergo detailed line segment comparison. This segmentation reduces the number of forms requiring detailed processing while maintaining recognition accuracy.
Solution Approach 2:
The patent performs preliminary coarse classification before detailed identification processing. By first classifying forms into candidate groups based on layout features and then performing detailed line segment correlation only on these candidates, the system prepares and filters data in advance, reducing the computational burden of subsequent detailed processing.
2Adaptability or versatility
If the number of line segments associated with a single form ID is increased, then recognition coverage is improved, but processing load increases due to more line segment pairs
Solution Approach 1:
The patent extracts and removes non-corresponding line segment pairs from the detailed identification process. By identifying and excluding line segment pairs that do not correspond between the input form and candidate forms, the system reduces the number of comparisons required while maintaining recognition coverage for forms with multiple line segments.
Solution Approach 2:
The patent performs partial detailed identification processing by focusing only on corresponding line segment pairs rather than all possible pairs. This partial action approach processes only the necessary subset of line segments that contribute to accurate form identification, avoiding unnecessary computational overhead from excessive comparisons.
3Measurement precision
If all line segment pairs are correlated for detailed identification, then recognition accuracy is maintained, but processing efficiency decreases
Solution Approach 1:
The patent performs preliminary filtering to identify corresponding line segment pairs before detailed correlation processing. By pre-identifying which line segments correspond between the input form and candidate forms based on spatial and structural relationships, the system prepares a reduced set of pairs for detailed processing, improving efficiency while maintaining accuracy.
Solution Approach 2:
The patent performs partial correlation processing by correlating only the corresponding line segment pairs identified in the preliminary filtering stage, rather than all possible pairs. This partial action maintains recognition accuracy by focusing on relevant comparisons while significantly improving processing efficiency by excluding unnecessary correlations.
Data Source
AI summary
A method includes: extracting a first line segment pair including a combination of line segments selected from a plurality of line segments included in an image of a form to be recognized; calculating a first feature amount which represents a relationship between the line segments in the extracted first line segment pair; extracting a candidate for a form identifier of the form to be recognized based at least on the calculated first feature amount and a second feature amount in line segments in a second line segment pair correlated with a form identifier which is registered in advance; extracting corresponding line segment pairs which include a line segment correlated with the candidate for the form identifier and a line segment of the form to be recognized; and specifying the form identifier of the form to be recognized based at least on the degree of overlapping.