Information Processing System for OCR Accuracy via Parameter Selection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing document collation systems face challenges in achieving high OCR recognition accuracy, particularly with complex backgrounds or watermarks, and require extensive processing time, as current methods do not adequately distinguish between text and background, leading to incomplete or inaccurate character recognition.
Innovation Solution
An information processing system that performs image processing and character recognition by analyzing and selecting optimal settings for multiple image processing parameters, reducing the number of combinations to be tested through a candidate selection process, and determining recommended settings for achieving higher recognition accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If extensive image processing parameter combinations are tested to achieve high OCR recognition accuracy, then recognition accuracy is improved, but processing time increases significantly
Solution Approach 1:
The system performs preliminary analysis of document features (watermarks, complex backgrounds, text density) before conducting full image processing. Based on this preliminary assessment, it pre-selects a reduced set of relevant image processing parameters to test, rather than exhaustively testing all possible parameter combinations. This preliminary action significantly reduces processing time while maintaining high recognition accuracy.
Solution Approach 2:
The image processing parameter selection is segmented into multiple stages: first, document type identification; second, feature-specific parameter selection; third, targeted parameter optimization. This segmentation allows the system to focus computational resources on the most relevant parameters for each document type, reducing overall processing time while maintaining accuracy.
2Measurement precision
If image processing settings are optimized for specific document features, then recognition accuracy is improved, but system complexity increases
Solution Approach 1:
The system dynamically adjusts image processing parameters based on real-time document feature detection. Rather than maintaining fixed complex settings for all document types, the system adapts parameters according to the specific characteristics of each document (watermark presence, background complexity, text density), simplifying the system architecture while improving accuracy.
Solution Approach 2:
The system changes processing parameters based on detected document features. For example, when watermarks are detected, specific denoising parameters are applied; when text density is high, contrast enhancement parameters are adjusted. This parameter-based adaptation allows the system to handle diverse document types without requiring complex structural changes.
3Measurement precision
If all image processing parameters are tested to ensure optimal settings, then recognition accuracy is improved, but the number of processing combinations increases
Solution Approach 1:
The system performs preliminary document analysis to identify key features (watermarks, backgrounds, text patterns) before parameter testing. Based on this preliminary information, it pre-filters the parameter combinations to test, focusing only on those relevant to the detected document type and features. This reduces the number of combinations from potentially thousands to a manageable subset.
Solution Approach 2:
Different image processing parameters are applied to different regions or aspects of the document based on local characteristics. For example, watermark regions receive different processing than text regions. This local quality approach reduces the overall number of parameter combinations needed, as each region is optimized independently with fewer parameters.
Data Source
AI summary
An information processing system includes circuitry. The circuitry performs image processing on an image to generate a processed image. The circuitry acquires a correct character string for a text area in the processed image. The circuitry acquires a recognized character string by performing character recognition on the text area in the processed image. The circuitry performs collation between the correct character string and the recognized character string. The circuitry displays one or more windows reflecting a result of the collation between the correct character string and the recognized character string on a display for allowing a user to evaluate whether the image processing is suitable for character recognition. The circuitry controls a displaying manner of at least one window of the one or more windows to vary according to the result of the collation.