Multi-mode document layout identification using aggregated similarity scores
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional approaches to identifying document layouts are either not specific enough to distinguish similar layouts or not robust enough to handle small variations, leading to inaccurate determinations in enterprise software applications processing scanned documents.
Innovation Solution
A system and method for multi-mode identification of document layouts that determines layout characteristics such as spatial positions and numeric representations of document features, generates an aggregated similarity score by comparing these characteristics to predefined layouts, and identifies the layout based on a threshold score, allowing for accurate document processing operations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional approaches are used to identify document layouts, then the process is simple, but the accuracy is insufficient to distinguish similar but different layouts
Solution Approach 1:
The patent segments layout identification into multiple independent characteristics: spatial position of document features, numeric representation of features, and template matching scores. Each characteristic is evaluated separately and then aggregated to form a comprehensive layout identification decision, improving accuracy without requiring a single complex identification system
Solution Approach 2:
The patent transforms layout identification from a single categorical label into a multi-parameter evaluation system. By converting layout characteristics into quantitative parameters (spatial coordinates, numeric representations, similarity scores) and applying threshold-based aggregation, the system achieves higher precision while maintaining manageable complexity through standardized parameter processing
2Reliability
If conventional approaches are used to identify document layouts, then the system is simple, but the robustness is insufficient to handle small variations in layout
Solution Approach 1:
The patent performs preliminary extraction and normalization of document features before layout identification. By pre-processing spatial positions and numeric representations of document features, and pre-defining multiple layout templates with their characteristic parameters, the system builds robustness against variations while keeping the identification logic itself relatively simple
Solution Approach 2:
The patent implements a feedback mechanism through aggregated similarity scoring. The system compares extracted document features against multiple predefined layout templates, accumulates similarity scores across different characteristics, and uses threshold-based feedback to confirm or reject layout identifications. This feedback loop enhances reliability by validating identifications through multiple independent checks
3Measurement precision
If multi-mode identification with multiple characteristics is used, then layout identification accuracy improves, but the processing time increases
Solution Approach 1:
The patent applies partial action by selectively evaluating layout characteristics based on confidence thresholds. The system calculates spatial positions and numeric representations for all documents, but only performs detailed template matching and aggregation for documents that meet preliminary criteria, reducing unnecessary processing while maintaining high accuracy for critical identifications
Solution Approach 2:
The patent segments the processing pipeline into independent, parallelizable stages: feature extraction, spatial position calculation, numeric representation generation, template matching, and score aggregation. This segmentation allows computationally intensive operations to be distributed and executed concurrently, mitigating the time cost of multi-mode identification
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A method is provided for multi-mode identification of document layouts. The method may include determining, based on a received document, a plurality of layout characteristics including a spatial position of one or more document features included in the received document and/or a numeric representation of the one or more document features included in the received document. The method may include generating an aggregated similarity score by at least comparing the plurality of layout characteristics to a first plurality of predefined layout characteristics of a first predefined layout of a plurality of predefined layouts. The method may further include identifying a layout of the received document as the first predefined layout of the plurality of predefined layouts based on the aggregated similarity score meeting a threshold score. The method may also include performing a document processing operation based on the identified layout. Related systems and methods are provided.