Scan Image Sectioning Accuracy via User-Corrected Candidates
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for dividing scan image data automatically identify sectioning positions based on user-specified page images, but may incorrectly divide data at unintended positions, leading to incorrect document segmentation.
Innovation Solution
An information processing system that displays a list of page images, identifies similar page images as sectioning position candidates, and allows users to correct these candidates before dividing the scan image data, ensuring accurate document segmentation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If automatic identification of sectioning positions is performed based on user-specified page images, then processing efficiency is improved, but sectioning accuracy deteriorates due to unintended divisions
Solution Approach 1:
The system displays automatically identified sectioning position candidates to the user and accepts user corrections. This feedback loop allows the system to learn from user corrections and improve future automatic identifications, resolving the contradiction between automatic processing efficiency and sectioning accuracy.
Solution Approach 2:
The system performs preliminary automatic identification of sectioning position candidates before final division, then presents these candidates to the user for verification and correction. This preliminary action separates the automatic processing stage from the final decision stage, maintaining both efficiency and accuracy.
2Speed
If automatic division is performed without user verification, then processing speed is improved, but reliability deteriorates due to incorrect document segmentation
Solution Approach 1:
The system performs automatic identification of sectioning position candidates (partial action) rather than complete automatic division, then requires user verification for the final decision. This partial automation maintains processing speed while ensuring reliability through user verification of the generated candidates.
3Measurement precision
If user correction of sectioning positions is allowed, then sectioning accuracy is improved, but device complexity increases due to additional interaction interfaces
Solution Approach 1:
The system uses the user's own corrections as training data to improve its automatic identification capability. The user's correction actions serve to retrain the learning model, allowing the system to become more accurate over time without requiring complex intervention interfaces for each operation.
Data Source
AI summary
The present disclosure helps prevent scan image data obtained by bulk reading of a plurality of documents from being divided at a position unintended by a user. An apparatus according to the present disclosure can display, on a screen, a list of a plurality of page images obtained by reading a plurality of documents in bulk, display, on the screen where the list of the plurality of page images is displayed, one or more page images similar to a page image specified by a user as one or more sectioning position candidates in a manner discriminable from other page images of the plurality of page images, and correct the one or more sectioning position candidates based on a correction instruction from the user.