Visual Query Document Retrieval via OCR Quality Scoring
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current systems for searching documents require users to input specific search terms, which can be cumbersome when trying to locate a specific document, and lack the ability to efficiently process visual queries to retrieve the canonical source document.
Innovation Solution
A computer-implemented method that processes visual queries by performing optical character recognition (OCR) on images of documents to identify high-quality text strings, allowing for the retrieval of the canonical source document matching the query and sending relevant portions to the client system.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If term-based searching is used, then search functionality is provided, but user convenience deteriorates due to requiring manual input of search terms
Solution Approach 1:
The patent replaces the mechanical process of manual text input with an optical system. Users capture images of documents using a camera or scanner, and the system automatically performs OCR (optical character recognition) to extract text from the images. This substitution eliminates the need for manual typing of search terms, significantly improving user convenience and reducing time loss.
2Measurement precision
If visual query processing is implemented, then document retrieval accuracy is improved, but system complexity increases
Solution Approach 1:
The patent segments the visual query processing into distinct functional modules: image capture module, OCR processing module, text extraction module, and document matching module. Each module handles a specific aspect of the processing pipeline, making the overall complex system more manageable and maintainable while achieving high document retrieval accuracy.
Solution Approach 2:
The patent introduces an intermediary OCR processing layer between the visual query input and the document retrieval system. This intermediary converts images into text, bridging the gap between visual input and text-based document search, thereby improving accuracy without directly complicating the core retrieval mechanism.
3Reliability
If high quality text extraction is performed, then search result quality is improved, but processing time increases
Solution Approach 1:
The patent applies local quality by focusing OCR processing only on relevant regions within the captured image. Instead of processing the entire image uniformly, the system identifies and processes only the regions containing text of interest, thereby maintaining high search result quality while reducing overall processing time.
Solution Approach 2:
The patent implements partial action by extracting only the necessary high-quality text strings required for document matching, rather than transcribing all text in the document. This selective extraction approach maintains reliability for search purposes while improving processing efficiency.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A server system receives a visual query from a client system. The visual query is an image containing text such as a picture of a document. At the receiving server or another server, optical character recognition (OCR) is performed on the visual query to produce text recognition data representing textual characters. Each character in a contiguous region of the visual query is individually scored according to its quality. The quality score of a respective character is influenced by the quality scores of neighboring or nearby characters. Using the scores, one or more high quality strings of characters are identified. Each high quality string has a plurality of high quality characters. A canonical source document matching the visual query that contains the one or more high quality textual strings is identified and retrieved. Then at least a portion of the canonical document is sent to the client system.