Document Layout Analysis Using Character Fully-Connected Graphs
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current document layout analysis methods lack accuracy in understanding document structure due to insufficient utilization of character information and spatial semantic features, leading to suboptimal recognition of text sequences and table relationships.
Innovation Solution
A method involving the generation of a character fully-connected graph based on character information, extraction of spatial semantic features, and classification of vertices and edges to accurately determine document layout, utilizing pre-trained models for vertex and edge type recognition.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional document layout analysis methods are used, then the analysis process is simple, but the accuracy of understanding document structure is insufficient
Solution Approach 1:
The patent segments the document layout analysis into multiple components: character-level information extraction, spatial relationship analysis, semantic feature extraction, and graph-based structure reconstruction. Each component processes specific aspects of the document independently before integrating results, thereby improving overall accuracy while maintaining manageable complexity through modular design
Solution Approach 2:
The patent transitions from traditional 2D image-based layout analysis to a multi-dimensional approach by extracting character information (position, size, orientation), spatial semantic features (relationships between characters), and constructing graph structures (nodes and edges). This dimensional expansion enables comprehensive capture of document structure that traditional methods miss
2Reliability
If character information and spatial semantic features are fully utilized, then the detection accuracy of text sequences and table relationships is improved, but the processing complexity increases
Solution Approach 1:
The patent performs preliminary extraction and processing of character information (position, size, orientation) and spatial semantic features before constructing the final layout analysis. By pre-processing and organizing this information into structured formats (graphs, feature vectors), the system reduces processing complexity during the main analysis phase while maintaining high detection accuracy
Solution Approach 2:
The patent introduces intermediate representations such as character graphs, spatial relationship graphs, and semantic feature vectors that mediate between raw character data and final layout analysis. These intermediaries organize and simplify the complex relationships between characters, enabling accurate detection of text sequences and table relationships without directly processing all raw data simultaneously
Data Source
AI summary
The disclosure may provide a method for obtaining a document layout, an electronic device, and a storage medium. The method may include: obtaining a plurality of characters and character information of each of the plurality of characters in a target document; generating a character fully-connected graph based on the plurality of characters and the character information of each of the plurality of characters; obtaining a spatial semantic feature of each of the plurality of characters; generating types of character vertices based on the character information of each of the plurality of characters; generating types of edges based on the spatial semantic feature of each of the plurality of characters, and the character information of each of the plurality of characters; and obtaining a document layout of the target document based on the types of the character vertices and the types of the edges.