Mobile Image Text Search with OCR Error Correction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image-based search methods on mobile devices are ineffective in handling poor-quality images, failing to detect and recognize text with errors, and are unable to search for content not pre-specified in the image database, especially when images contain logos, alphanumeric characters in various languages, and background variations.
Innovation Solution
A method and system that automatically detects and recognizes text in images captured by mobile devices, processes text geometry information, and searches for matches across domain databases, eliminating the need for manual text editing or writing, using an image processing module and matching engine with OCR and N-gram algorithms to improve image quality and accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If image-based search is used to search for information corresponding to an image captured by a mobile communications device, then the search can be performed automatically without manual text input, but the search accuracy deteriorates due to poor image quality and text extraction errors
Solution Approach 1:
The patent introduces an intermediary text correction system that acts as a mediator between the OCR extraction and the actual search process. This system uses multiple strategies including typo correction, synonym expansion, and fuzzy matching to bridge the gap between extracted text and database queries, thereby maintaining automatic search convenience while improving accuracy despite poor image quality
Solution Approach 2:
The patent applies preliminary action by performing text correction and expansion before the actual search operation. The system pre-processes the extracted text by correcting common errors, expanding abbreviations, and generating alternative spellings in advance, so that when the search is executed, the query is already optimized for better matching against the database
2Measurement precision
If the search is made robust to handle text extraction errors and long queries, then the search accuracy improves, but the search time increases
Solution Approach 1:
The patent segments the search process into multiple independent stages: text extraction, error correction, query expansion, and actual searching. By dividing the robust search process into discrete segments, the system can apply optimizations at each stage and parallelize certain operations, maintaining high accuracy while reducing overall search time
Solution Approach 2:
The patent implements partial action by applying text correction and expansion selectively rather than to all queries uniformly. The system identifies which parts of the query need correction based on confidence scores and error patterns, applying processing only where necessary, thus achieving robust accuracy without the full time cost of processing every aspect of every query
3Device complexity
If image-based search only searches for exact matches corresponding to the image, then the search process is simple, but valuable information such as company name or address in the textual part is not found
Solution Approach 1:
The patent makes the search process dynamic by automatically adapting the search strategy based on the image content and extracted text. The system dynamically switches between exact match searching and text-based content searching, and adjusts the level of query expansion and correction applied, thereby finding textual information like company names and addresses without requiring a completely complex predetermined process
4Adaptability or versatility
If the mobile communications device camera captures images, then the user can perform image-based search, but the images are poor in quality with poor contrast, poor resolution, blurring, and intensity variations
Solution Approach 1:
The patent converts the harmful effect of poor image quality into a benefit by developing specialized text extraction and correction algorithms that are specifically designed to handle low-quality images. Instead of requiring high-quality images, the system learns from and adapts to the characteristic errors introduced by mobile device cameras, turning the limitation into a manageable condition with dedicated processing strategies
Data Source
AI summary
Present invention relates to a method and system for automatic searching for information on a network in response to an image query sent by a user. The image query includes an image that is captured by using a mobile communications device with a camera. The image is processed to detect the text present in it. The detected text is then recognized using an OCR. Subsequently, the text is searched for matches in the corresponding domain database, selected from the various domain databases present in the network. Thereafter, selected matches and additional related information is sent to the user.