Mixed Content OCR Image Processing Apparatus
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing OCR systems face challenges in accurately recognizing both printed and handwritten characters in mixed content documents, as the accuracy of character recognition deteriorates when both types of characters coexist, and existing techniques require a registered dictionary mechanism that limits compatibility with new OCR engines.
Innovation Solution
An image processing apparatus that extracts and clips areas of handwritten and printed characters, generating combined images for character recognition by associating each other, allowing for improved recognition accuracy and compatibility with various OCR engines without the need for a registered dictionary.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If both printed and handwritten characters are recognized together in mixed content documents, then complete information extraction is achieved, but character recognition accuracy deteriorates
Solution Approach 1:
The document image is segmented into printed character regions and handwritten character regions using image processing techniques. By separating the recognition of printed and handwritten characters into distinct processing streams, the system maintains high recognition accuracy for both types while handling mixed content documents effectively.
Solution Approach 2:
Image processing techniques serve as an intermediary step between document scanning and OCR recognition. This intermediary processing identifies and separates printed and handwritten character areas, allowing each type to be recognized by the most appropriate OCR engine, thereby maintaining high accuracy while handling mixed content.
2Measurement precision
If a registered dictionary mechanism is used to improve handwritten character recognition, then recognition accuracy is enhanced, but compatibility with new OCR engines is limited
Solution Approach 1:
The system extracts only the handwritten character image data from the document, separating it from printed characters. This extracted handwritten data is then processed by OCR engines without requiring the complex registered dictionary mechanism, enabling compatibility with various OCR engines while maintaining recognition accuracy through proper image preparation and processing.
Solution Approach 2:
The system creates a simplified copy of the handwritten character region with optimized image properties suitable for OCR processing. This copied and prepared image data can be processed by different OCR engines without requiring them to implement specific registered dictionary mechanisms, thereby improving engine compatibility while maintaining accuracy.
3Adaptability or versatility
If only handwritten characters are extracted for recognition, then OCR engine compatibility is improved, but recognition accuracy cannot be enhanced through natural language processing
Solution Approach 1:
The system segments the document into printed and handwritten regions, then processes each segment appropriately. Printed character regions are processed using OCR engines that can leverage natural language processing and contextual information, while handwritten regions are processed with optimized image preparation. This segmentation enables both compatibility and accuracy enhancement simultaneously.
Solution Approach 2:
The system merges the results of printed character recognition and handwritten character recognition into a unified output. By combining the strengths of both recognition approaches and using the contextual relationship between printed and handwritten characters, the system achieves both OCR engine compatibility and enhanced recognition accuracy through natural language processing.
Data Source
AI summary
An image processing apparatus that generates an image for character recognition from a read image includes at least one memory that stores instructions, and at least one processor that executes the instructions to perform extracting of an area of handwritten character information and an area of printed character information from the read image, clipping of a partial image of the area of handwritten character information and a partial image of the area of printed character information out of the read image, and generating of the image for character recognition by combining the partial image of the area of handwritten character information and the partial image of the area of printed character information being associated with each other.