Character Recognition TRIZ Cases
- Secure Document Data Capture System with Redacted Exception Images
A secure document data capture system generates exception images with redacted context portions for suspect character correction.
- Dynamic Zone Mapping for Document Text Extraction
Automated zone detection eliminates manual drawing errors by using indexed text data to map occurrences directly to fields.
- OCR Error Correction via Font-Specific Error Models
A font-specific error model corrects systematic kerning and weight errors in OCR text using the Wagner-Fischer algorithm.
- Video Text Region Preprocessing for OCR Accuracy
Locates and enhances text images in video frames to resolve inefficient region determination and improve optical character recognition accuracy.
- License Plate Localization via Texture and Connected Component Analysis
A license plate localization method combines top-down texture analysis with bottom-up connected component processing to identify plate regions in vehicle images.
- Invisible Junctions for Document Recognition
Invisible junctions extract intrinsic skeletons from noisy images to resolve recognition accuracy and stability trade-offs in document retrieval.
- Automatic Graph Data Extraction Using Deep Learning
Automated graph data extraction replaces manual labeling with deep learning to improve precision and speed.
- Image Forming Apparatus Document Classification Automation
Automated document classification reduces manual sorting errors by generating guide information through optical character recognition.
- Image Data Masking via Classification-Based Category Search
An information processing apparatus identifies image data classifications to search for specific character strings within associated categories.
- Character Recognition Region Partitioning for Abnormal Layouts
A partitioning unit analyzes character size distributions to re-segment specific image regions containing out-of-range value characters.
- Label Detection Using Special Marks for OCR Accuracy
A label detection system locates product labels using special marks to isolate text regions from complex image backgrounds.
- Character Normalization via Adaptive Margin Bounding Boxes
A normalization apparatus adds margins to character bounding boxes based on adjacent character width data.
- Image Processing Apparatus Folder Naming via OCR Selection
An image processing apparatus selects a folder name from recognized text to store multi-page scan data.
- Business Card Array Detection via Character Distribution Analysis
A detection unit identifies business card arrays by analyzing character string distribution and size patterns within scanned document images.
- Character Recognition Device Dynamic Width Estimation
A character recognition device estimates standard character width from image data to handle mixed character types.
- OCR Word Fragment Merging via Normalized Distance and Predefined Expressions
A computer system merges adjacent word fragments in OCR data using normalized horizontal distances and predefined expressions.
- Gesture Recognition Using Neural Networks on Resistive Touch Panels
Neural networks calculate angles between sequential touch points to recognize gestures, enabling multi-touch simulation on low-cost resistive panels.
- Document Scanning Control Device for Adaptive File Format Selection
A control device detects table structures in scanned images to generate appropriate spreadsheet or word processor files.
- Text Rectifier Using TILT for Deformed Image OCR
A Text Rectifier applies rank minimization to recover and remove image deformations from text regions.
- Facial Recognition System for Automated Event Execution
A facial recognition system extracts identity data from images to automate event execution workflows.
- Dynamic OCR Selection for Document Processing Efficiency
A selection section chooses between multiple conversion processing sections to optimize character code data generation.
- Character Image Extraction Using Overlap Merging and Size Filtering
A character image extracting apparatus merges overlapping section elements to isolate complete glyphs from linked components.
- Multi-modal Document Feature Extraction via Hierarchical Tree
Hierarchical feature trees combine text and image extractions into input vectors, reducing training data requirements for rare document types.
- Document Image Note Masking and File Naming
Image processing apparatus extracts document areas and note regions to generate electronic files with automatic file names.
- Text Detection via Neural Network Feature Matrix Clustering
Fully connected layer mapping and text clustering algorithm resolve dynamic boundary inaccuracies in character positioning.