Text Block Recognition via Discrete Character Semantic Connectivity
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current text block recognition methods face challenges in accurately identifying and reconstructing table content, especially when border lines are inconspicuous or absent, and in handling vertical text, leading to inconsistent cell segmentation and recognition accuracy.
Innovation Solution
The approach employs discrete character recognition using semantic connectivity analysis, leveraging pre-trained language models like BERT to organize characters into meaningful content, and reconstructs tables based on character positions, outputting the text blocks in formats like XML or CSV.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional text block recognition methods are used, then processing is simpler, but recognition accuracy deteriorates when border lines are inconspicuous or absent
Solution Approach 1:
The patent segments the text recognition process into discrete character detection, position identification, and semantic connectivity analysis stages. By dividing the document into individual character units and analyzing their spatial relationships, the method achieves accurate text block recognition without relying on border lines, thus improving recognition accuracy while maintaining manageable complexity through structured processing steps.
Solution Approach 2:
The patent introduces semantic connectivity analysis as an intermediary mechanism between character detection and text block formation. This intermediary step analyzes the semantic relationships and spatial positions of detected characters to determine their connectivity, enabling accurate text block identification even when traditional visual boundaries are absent or inconspicuous.
2Measurement precision
If discrete character recognition with semantic connectivity is used, then recognition accuracy improves, but processing complexity increases
Solution Approach 1:
The method segments text processing into discrete character-level operations followed by semantic connectivity analysis. This segmentation allows the system to handle complex semantic relationships through modular processing steps, improving accuracy while keeping each processing stage manageable and independent.
Solution Approach 2:
The patent performs semantic connectivity analysis at the character level, which is more fine-grained than traditional word or line-level processing. This partial action approach analyzes connectivity between individual characters rather than entire text blocks, enabling higher precision in determining text block boundaries while the modular nature of character-level processing keeps computational complexity manageable.
3Manufacturing precision
If character position analysis is performed, then text block organization improves, but processing time increases
Solution Approach 1:
The patent performs character detection and position identification as preliminary actions before semantic connectivity analysis. By pre-processing and storing character positions, the system avoids redundant calculations during the connectivity analysis phase, thereby improving text block organization precision while minimizing additional processing time through efficient use of pre-computed position data.
Data Source
AI summary
In an approach for a text block recognition in a document, a processor detects characters in the document using an object detection technique. A processor identifies positions of the detected characters in the document. A processor analyzes semantic connectivity among the detected characters based on the positions and semantic connectivity of the characters. A processor recognizes text blocks of related characters based on the semantic connectivity analysis. A processor outputs the text blocks associated with the related characters.