Document Image Cleansing via Adaptive Cleaning and OCR Selection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image processing technologies face challenges in accurately generating image data that does not represent information of a deletion target, leading to reduced recognition accuracy of characters in image data, particularly when using fixed processing methods regardless of the information present.
Innovation Solution
An information processing apparatus that utilizes a processor to perform cleansing processing based on the appearance characteristics of documents, generating second image data that excludes information of a deletion target while retaining relevant information, using cleansing learning devices and character recognition learning devices implemented with artificial intelligence to enhance accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If fixed processing is applied to generate image data regardless of information represented in the input image, then the processing is simple and fast, but the accuracy of generating image data not representing information of a deletion target deteriorates
Solution Approach 1:
The patent implements dynamic processing by selecting different cleaning methods based on document type detection. The system determines whether to apply first cleaning (removing backgrounds) or second cleaning (removing specific elements like seals) based on the detected document type, making the processing adaptive rather than fixed. This resolves the contradiction by maintaining simplicity for common cases while enabling enhanced accuracy when needed.
Solution Approach 2:
The system changes processing parameters based on document characteristics. By detecting document types (e.g., certificates, forms) and adjusting the cleaning approach accordingly, the system optimizes the balance between processing complexity and output accuracy. Different document types trigger different processing intensities and methods.
2Measurement precision
If fixed recognition processing is applied regardless of information represented in image data, then the processing is simple and fast, but the accuracy of character recognition deteriorates
Solution Approach 1:
The patent implements dynamic character recognition by selecting different OCR engines based on document type. After detecting the document type through cleaning processing, the system chooses appropriate recognition methods tailored to that document type, thereby improving recognition accuracy without applying complex processing to all documents uniformly.
Solution Approach 2:
The system applies different recognition qualities to different regions or types of documents. By identifying specific document types and applying specialized recognition processing only where needed, the system enhances local recognition accuracy while maintaining overall system efficiency.
3Productivity
If image data representing information of a deletion target is generated, then the processing is simple, but the accuracy of subsequent character recognition deteriorates
Solution Approach 1:
The patent applies preliminary cleaning processing to remove information of deletion targets (such as backgrounds, seals, or irrelevant elements) before character recognition. This preliminary action ensures that the main processing stage receives optimized input data, improving recognition accuracy without significantly impacting overall processing efficiency.
Solution Approach 2:
The system extracts and removes specific elements that should not be recognized (deletion targets) from the image data before recognition processing. By taking out these interfering elements selectively based on document type, the system maintains processing efficiency while enhancing recognition accuracy.
Data Source
AI summary
An information processing apparatus includes a processor. The processor is configured to receive first image data representing a document, and generate, by processing corresponding to appearance characteristics of the document, second image data not representing information of a deletion target out of information represented in the first image data but representing information other than the information of the deletion target.