OCR Image Orientation Correction via Rotated Quality Scoring
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing OCR systems face significant accuracy issues when processing improperly oriented images, leading to diminished recognition performance and reduced reliability in extracting machine-readable text from medical charts and other documents.
Innovation Solution
A computer-implemented method that performs image orientation analysis using machine-readable text metadata, involving an OCR process, rotation analysis, and machine learning models to generate and evaluate rotated image data objects, ensuring high-quality OCR output by identifying optimal image rotation for accurate text extraction.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If OCR processing is performed on improperly oriented images, then processing speed is maintained, but OCR accuracy significantly deteriorates
Solution Approach 1:
The system performs preliminary image orientation correction by generating rotated versions of the input image and evaluating their OCR quality scores before final text extraction. This preliminary action ensures that the OCR processing is performed on properly oriented images, thereby improving accuracy without adding significant complexity to the overall system.
2Measurement precision
If multiple rotated image versions are generated and evaluated, then OCR accuracy improves, but processing time increases
Solution Approach 1:
The system generates a limited set of rotated image versions (typically 0°, 90°, 180°, 270°) rather than exhaustively trying all possible orientations. This partial action approach is sufficient to correct common orientation errors while avoiding excessive processing time that would result from more comprehensive orientation analysis.
Solution Approach 2:
The system uses quality scores generated by machine learning models to evaluate each rotated image version and provides feedback to select the optimal orientation. This feedback mechanism allows the system to efficiently identify the best-oriented image without manually inspecting all rotated versions, thereby reducing processing time while maintaining high accuracy.
3Measurement precision
If quality scoring using machine learning models is applied to all images, then text quality assessment improves, but computational resources increase
Solution Approach 1:
The system applies machine learning-based quality scoring selectively to specific regions or aspects of the OCR output rather than uniformly to all images. By focusing computational resources on evaluating the most critical quality aspects, the system achieves accurate quality assessment while reducing overall computational resource consumption.
Data Source
AI summary
Systems and methods are configured for correcting the orientation of an image data object subject to optical character recognition (OCR) by receiving an original image data object, generating initial machine readable text for the original image data object via OCR, generating an initial quality score for the initial machine readable text via machine-learning models, determining whether the initial quality score satisfies quality criteria, upon determining that the initial quality score does not satisfy the quality criteria, generating a plurality of rotated image data objects each comprising the original image data object rotated to a different rotational position, generating a rotated machine readable text data object for each of the plurality of rotated image data objects and generating a rotated quality score for each of the plurality of rotated machine readable text data objects, and determining that one of the plurality of rotated quality scores satisfies the quality criteria.