OCR Apparatus Iterative Region-Specific Scanning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional Optical Character Recognition (OCR) methods fail to accurately extract data from articles due to low resolution, small-sized data, and skewed regions, as they perform a single scanning process applicable to both textual and image regions, leading to incomplete data conversion.
Innovation Solution
A method and apparatus for OCR that acquire images, identify textual regions, and perform OCR using predetermined settings, iteratively refining the OCR process based on quality parameters to ensure accurate extraction of textual data from articles.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If a single scanning process is used for entire article, then the scanning speed is fast, but the data extraction completeness deteriorates
Solution Approach 1:
The article is divided into multiple regions of interest (ROIs) based on detected features such as text density, image quality, and character clarity. Each ROI is then scanned separately with optimized parameters tailored to its specific characteristics, allowing both fast processing of simple regions and thorough extraction of difficult regions.
Solution Approach 2:
The scanning process dynamically adjusts parameters such as resolution, scan depth, and processing intensity based on the characteristics of each detected ROI. This dynamic adaptation enables the system to maintain high speed for straightforward regions while applying more rigorous scanning only where necessary.
2Measurement precision
If high resolution focus is used, then the OCR accuracy is improved, but the processing time increases
Solution Approach 1:
Different resolution levels and processing intensities are applied to different regions of the article based on their specific requirements. Regions with clear, large text receive standard processing, while regions with small, skewed, or low-contrast characters receive enhanced resolution and iterative scanning to ensure accurate OCR.
Solution Approach 2:
The system performs partial high-resolution scanning only on specific regions that require it, rather than applying high resolution uniformly across the entire article. This selective approach maintains OCR accuracy for difficult regions while minimizing the overall processing time burden.
3Device complexity
If conventional scanning is performed, then the process is simple, but small sized data and skewed regions are not recognized
Solution Approach 1:
Before performing OCR, the system performs preliminary analysis to detect and identify regions containing small-sized or skewed data. These challenging regions are then subjected to specialized preprocessing operations such as deskewing, magnification, and contrast enhancement, ensuring they are properly prepared for accurate recognition.
Solution Approach 2:
The scanning and processing parameters are dynamically changed based on the detected characteristics of each ROI. For small-sized data, the system increases magnification and adjusts sampling rates. For skewed regions, it modifies the scanning angle and applies corrective transformations, all automatically adapted to maintain process simplicity while improving recognition accuracy.
Data Source
AI summary
Embodiments of the present disclosure disclose a method for performing Optical Character Recognition (OCR) of an article. The method comprises acquiring an image of the article. The image of the article is scanned using predetermined scan settings. Then, textual regions of the scanned image of the article are identified. The OCR of the at least one of the textual regions is performed using predetermined OCR settings. One or more textual regions of the textual regions are marked upon determining an error in performing the OCR of the one or more textual regions. The OCR of the one or more textual regions is iterated as per one or more predefined OCR scanning parameters based on an OCR quality of the one or more textual regions upon marking the one or more textual regions.