Borderless Table Extraction via Binary Image Gap Clustering
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing Optical Character Recognition (OCR) techniques struggle to accurately extract textual data from documents with varying layouts and structures, particularly those without defined borders or clearly separated structure cells.
Innovation Solution
A method involving the conversion of documents into binary images, followed by the creation of text character regions and text line regions using morphological operations. Gap regions between these regions are identified, clustered, and used to enclose regions of interest with horizontal and vertical structure lines, ultimately generating a list of cell coordinates.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If generic algorithms are used to detect and extract information from documents with varying layouts, then the algorithm can be broadly applicable, but the accuracy and capability to handle borderless structures deteriorates
Solution Approach 1:
The patent segments the document processing into distinct stages: converting to binary image, creating text character regions, merging into text line regions, identifying gap regions, clustering gaps, and enclosing with structure lines. This segmentation allows each stage to be optimized independently, enabling accurate detection of borderless structures while maintaining broad applicability.
Solution Approach 2:
The patent performs preliminary actions by first converting the document to a binary image and pre-defining the inter-text region distance before actually detecting the structure. This preliminary processing simplifies subsequent detection tasks and improves accuracy by preparing the data in an optimized format.
2Device complexity
If OCR techniques assume standard document layouts, then the processing can be simplified, but the capability to handle varying and borderless structures deteriorates
Solution Approach 1:
The patent employs dynamic processing by adaptively determining structure lines based on gap region clustering and text line region coordinates, rather than assuming fixed standard layouts. The algorithm dynamically adjusts to the actual document structure, maintaining simplicity while enhancing adaptability to varying layouts.
Solution Approach 2:
The patent changes parameters during processing by using pre-defined inter-text region distance and dynamically calculating structure line positions based on gap region characteristics. This allows the system to maintain simple processing while adapting to different document structures through parameter adjustment.
3Adaptability or versatility
If heuristics are used to detect structure in documents, then the method can be more flexible, but the accuracy of extraction deteriorates
Solution Approach 1:
The patent replaces heuristic-based mechanical detection with a systematic image processing approach using binary conversion, morphological operations, and mathematical clustering. This substitution maintains flexibility while significantly improving accuracy by using rigorous computational methods instead of rule-of-thumb heuristics.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
This approach enables the effective extraction of borderless structures from documents, allowing for the accurate identification and formatting of textual information into structured form for further processing.
Implementation Method 1
the plurality of text character regions are generated by connecting one or more consecutive text characters within a text line of the plurality of text lines, using at least one morphological operation
Implementation Method 2
The document received may be converted into a binary image, wherein the document comprises a plurality of text characters in a plurality of text lines. Converting may include changing the color of the background to black color and changing the color of the foreground to white color
Data Source
AI summary
A method and system of extracting borderless structure using image processing is disclosed. The method may include converting a received document into a binary image comprising a plurality of text characters. A first image is created comprising a plurality of text blobs by connecting text characters and merging the plurality of text blobs to create one or more text line blobs to generate a second image. Further the first image and the second image are compared to generate a third image comprising a plurality of gap blobs. The gap blobs are clustered into one or more groups to determine a localized region of interest (ROI). Further lines are identified within the ROI using pixel density and separated into rows and columns. The final output contains list of cell coordinates.