Document Area Division in Multi-Cropping Scanning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In multi-cropping processing, accurately recognizing document areas from scanned images is challenging when documents are not separated sufficiently, leading to difficulties in determining individual document areas, requiring users to re-scan and rearrange documents.
Innovation Solution
An image processing system with a user interface that displays a preview screen showing the results of document recognition, allowing users to easily modify and divide document areas by providing buttons to adjust the recognition state, including features like division prediction and adjustment of document areas based on object detection and edge extraction processing.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Extent of automation
If document recognition processing is performed based on distance between objects, then document areas can be determined automatically, but accuracy deteriorates when documents are placed closely together
Solution Approach 1:
The patent divides the document area determination process into multiple stages: initial automatic detection using object distance, followed by user review and manual adjustment if needed. This segmentation allows the system to automate the easy cases while providing manual intervention for difficult cases where documents are closely placed, thereby maintaining both automation and accuracy.
Solution Approach 2:
The patent implements a feedback mechanism where the system displays detected document areas to users for verification. Users can provide feedback by adjusting boundaries or re-detecting areas, and this feedback is used to improve the accuracy of subsequent detections. This closed-loop approach resolves the contradiction by allowing automatic processing to proceed while providing correction pathways when accuracy suffers.
2Measurement precision
If users rearrange documents to improve recognition accuracy, then document area detection accuracy improves, but time consumption increases
Solution Approach 1:
The patent performs preliminary document area detection automatically before requiring user intervention. By initially detecting document areas using object distance and presenting results to users, the system allows most cases to be processed without rearrangement. Users only need to intervene when the preliminary detection is insufficient, significantly reducing the time loss compared to requiring manual rearrangement for all cases.
Solution Approach 2:
The patent enables the system to self-correct detection errors by allowing users to adjust document area boundaries directly on the displayed image. Instead of requiring physical rearrangement and re-scanning, users can digitally modify the detected areas through the interface, and the system automatically processes these adjustments. This self-service approach eliminates the time-consuming rearrangement step while maintaining accuracy improvement.
3Productivity
If multiple documents are scanned together, then productivity increases, but document area recognition accuracy deteriorates when documents are not sufficiently separated
Solution Approach 1:
The patent segments the multi-document processing into automatic detection phase and user verification phase. During automatic detection, the system processes all documents together to maintain productivity. When detection accuracy is compromised due to close placement, the system selectively flags only those specific document areas requiring user review, rather than requiring manual processing of all documents. This segmented approach preserves productivity while improving accuracy for problematic cases.
Solution Approach 2:
The patent dynamically adjusts detection parameters based on the detected scene. When documents are detected to be closely placed, the system changes parameters such as object distance thresholds or detection sensitivity to improve individual document area recognition. This parameter adaptation allows the system to maintain high productivity in multi-document scanning while adjusting to maintain accuracy when documents are not sufficiently separated.
Data Source
AI summary
To make it possible for a user to easily modify the recognition state of a document by document recognition processing at the time of multi-cropping processing. A preview screen is displayed on a user interface, which displays the results of the document recognition processing for a scanned image obtained by scanning a plurality of documents en bloc on the scanned image in an overlapping manner. Then, a button for dividing the detected document area is displayed on the preview screen so that it is made possible for a user to easily perform division.