Scanned Book Data Processing via Character Segmentation and Aggregation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for processing scanned book data are inefficient in rearranging the layout and reducing data storage requirements, as they do not allow for advanced applications beyond simple text copying and fail to effectively compress data.
Innovation Solution
A method and apparatus that segment and recognize page image data to obtain precise locations of characters, aggregate character lines, classify character images, assign unique numbers, and compress data, enabling re-typesetting and reducing storage needs by removing character images before compression.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If a transparent layer with OCR words is overlaid on scanned images to enable text copying, then text copying capability is improved, but layout re-arrangement capability remains insufficient
Solution Approach 1:
The patent segments the scanned page into multiple character-level units with bounding boxes, extracting individual character information (position, content, visual appearance) to enable independent manipulation. This segmentation allows text to be copied while preserving the ability to re-arrange characters and lines freely, resolving the contradiction between easy text copying and layout re-arrangement capability.
Solution Approach 2:
The patent transitions from a two-dimensional scanned image to a multi-dimensional data structure that includes spatial information (bounding box coordinates), textual information (OCR recognition results), and visual information (character images). This dimensional expansion enables both text copying and flexible layout re-arrangement by manipulating data in multiple dimensions simultaneously.
2Measurement precision
If high DPI scanned images are used to preserve original page structure, then image quality is improved, but data storage and transmission becomes challenging
Solution Approach 1:
The patent extracts essential information from the scanned image (character positions, text content, visual appearance) and stores only this extracted data rather than the entire high-resolution image. By taking out only the necessary character-level information, the system maintains image quality for reference while dramatically reducing storage requirements for processing and transmission.
Solution Approach 2:
The patent creates a copy of the scanned page that includes both the original image data and extracted character information in a structured format. This copy enables text copying and layout re-arrangement operations while the original high-resolution scan serves as a reference, avoiding the need to store multiple full-resolution copies.
3Reliability
If character images are retained in the scanned book data, then original document restoration is possible, but data compression efficiency is reduced
Solution Approach 1:
The patent performs preliminary extraction and classification of character images before compression. Characters are grouped by visual appearance and assigned to categories, with only representative images stored for each category. This preliminary organization enables efficient compression while maintaining the ability to restore the original document by reconstructing characters from their categories.
Solution Approach 2:
The patent changes the representation parameters of character images from individual pixel data to categorical labels and representative samples. Instead of storing every character image in full detail, the system stores compressed categorical information that can be expanded back to the original characters, achieving both compression efficiency and restoration capability.
Data Source
AI summary
A method for processing data of a scanned book having a plurality of pages is disclosed. The method includes obtaining page image data from a page. The method further includes segmenting and recognizing the page image data to obtain locations of rectangular boxes corresponding to the respective characters and text codes for the respective characters. The method also includes obtaining respective aggregated character line information for each line of characters. The method further includes adjusting the rectangular boxes in accordance with the obtained aggregated character line information.