Image Processing Apparatus Overlap Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing multi-cropping processing techniques fail to accurately extract value information from scanned images when documents overlap, leading to incorrect registration and requiring users to re-scan documents, which is time-consuming and burdensome.
Innovation Solution
An image processing system that analyzes scanned images, identifies overlapping documents, and prompts users to re-scan specific documents where value information extraction has failed, using multi-cropping processing, OCR, and shielding determination to determine hidden character strings.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If multi-cropping processing is performed on scanned images, then document images can be extracted for further processing, but incorrect extraction occurs when documents overlap
Solution Approach 1:
The system performs preliminary analysis processing before final extraction to detect potential extraction failures. By analyzing the scanned image in advance and identifying documents with extraction failures, the system can prompt users to re-scan problematic documents before they are permanently registered, thus preventing incorrect data entry while maintaining efficient automated processing for successful extractions.
2Measurement precision
If users are prompted to re-scan all documents, then extraction accuracy can be improved, but user time and effort increase significantly
Solution Approach 1:
Instead of treating all documents uniformly, the system applies different processing strategies to different documents based on their extraction results. Documents with successful extraction are automatically registered without user intervention, while only documents with extraction failures are flagged for user review and re-scan. This localized approach ensures high accuracy for problematic documents while maintaining efficiency for the majority of successfully extracted documents.
3Productivity
If automated division is performed on overlapping documents, then processing speed increases, but value information extraction fails when important portions are hidden
Solution Approach 1:
The system implements a feedback mechanism where the results of analysis processing are used to determine subsequent actions. After attempting automated extraction from divided document images, the system evaluates whether extraction was successful. When extraction failure is detected, the system provides feedback to the user through a prompt, requesting re-scan of the specific document. This feedback loop ensures that no value information is lost due to overlapping, while maintaining high productivity for cases where automated processing succeeds.
Data Source
AI summary
At the time of registering an image in units of documents, which is obtained by performing multi-cropping processing for a scanned image, and value information extracted from each image in a database in association with each other, there is a case where it is not possible to extract the value information successfully because of an overlap between documents during a scan. In a case where this fact is noticed after registration, it becomes necessary for a user to perform the entire work from the beginning. In a case where there is a possibility that the value information is hidden by an overlap between documents during a scan or the like, a display is produced for prompting a user to scan the document again.