Partial Region Accuracy Feedback for OCR Preprocessing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current OCR processing systems pass entire image data for preprocessing, which can lead to decreased accuracy in downstream processing due to regions with low certainty degrees, potentially affecting important information.
Innovation Solution
An information processing apparatus that acquires image data, calculates accuracy for each partial region, and notifies preprocessing to focus on improving regions with low certainty degrees, allowing for targeted re-processing.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If cleansing processing is executed in units of entire images, then the processing is simple and fast, but the accuracy of downstream processing decreases in regions with low certainty degrees
Solution Approach 1:
The patent divides the image processing into two levels: entire image processing for overall quality assessment and partial region processing for localized accuracy improvement. The certainty degree calculation is performed for each partial region, allowing targeted re-processing only where needed, thus maintaining high processing speed while improving accuracy in critical areas.
Solution Approach 2:
The patent applies different processing strategies to different regions based on their certainty degrees. Regions with high certainty degrees are passed through without re-processing, while regions with low certainty degrees undergo targeted cleansing processing. This local differentiation optimizes both processing efficiency and accuracy.
2Device complexity
If entire image data is passed to downstream processing, then the processing flow is simple, but important information may be lost due to low certainty degree regions
Solution Approach 1:
The patent implements a feedback mechanism where the certainty degree of each partial region is calculated and used to determine whether re-processing is needed. This feedback loop ensures that only regions with low certainty degrees are re-processed, maintaining information quality while avoiding unnecessary processing complexity.
Solution Approach 2:
Instead of re-processing entire images, the patent applies partial re-processing only to specific regions with low certainty degrees. This partial action approach maintains information quality for critical regions while avoiding the complexity and resource consumption of full-image re-processing.
3Measurement precision
If high resolution image acquisition is used, then the certainty degree of OCR processing is improved, but the data size and processing time increase
Solution Approach 1:
The patent segments the image into multiple partial regions and calculates certainty degrees for each region independently. This allows selective re-processing of only those regions that require improvement, reducing the overall processing time compared to re-processing entire high-resolution images.
Solution Approach 2:
The patent applies processing effort selectively to partial regions rather than uniformly across the entire image. By identifying and re-processing only regions with low certainty degrees, the system achieves high OCR accuracy without the time penalty of processing all image data at maximum resolution.
Data Source
AI summary
An information processing apparatus includes a processor configured to acquire image data from preprocessing, calculate information related to accuracy of a result of processing for each partial region of an image corresponding to the acquired image data, and notify the preprocessing of the calculated information related to the accuracy and information for specifying a corresponding partial region.