Text Entity Recognition in Images via OCR and Pattern Matching
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional methods for identifying and processing text entities in real-world images, such as phone numbers and URLs, are prone to errors and inefficient, especially when the text is displayed on signs or boards at a distance from the computing device.
Innovation Solution
A system that uses optical character recognition (OCR) combined with heuristic tests and pattern matching algorithms to recognize and classify text entities in images, rendering overlays on the device's interface for user interaction, allowing functions like dialing or browsing associated with the recognized text.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of time
If manual input methods are used for phone numbers and URLs, then input accuracy can be maintained, but user time and effort increase significantly
Solution Approach 1:
The patent replaces manual mechanical input (typing phone numbers and URLs) with an automated optical recognition system. The camera captures images containing text entities, and OCR technology automatically extracts and processes the text, substituting the mechanical typing process with an automated visual recognition system that significantly reduces user time and effort.
Solution Approach 2:
The system enables self-service by allowing the device to automatically capture, recognize, and process text entities from images without requiring manual user input. The camera and OCR system work together to autonomously extract phone numbers, URLs, and other text entities, making the device serve itself in the information extraction task.
2Productivity
If OCR is used to recognize text from distant signs, then text extraction speed improves, but recognition accuracy deteriorates due to image quality issues
Solution Approach 1:
The patent applies preliminary action by performing image preprocessing operations before OCR recognition. The system enhances image quality through preprocessing steps such as noise reduction, contrast adjustment, and text region enhancement, preparing the image in advance for more accurate OCR recognition of distant or low-quality text.
Solution Approach 2:
The patent introduces an intermediary processing layer between image capture and OCR recognition. This intermediary layer includes image enhancement and text detection algorithms that bridge the gap between raw images of distant signs and the OCR engine, improving recognition accuracy by preprocessing the visual data before text extraction.
3Adaptability or versatility
If multiple text entity types are recognized, then system versatility improves, but processing complexity increases
Solution Approach 1:
The patent applies segmentation by dividing the text recognition process into distinct modules for different entity types (phone numbers, URLs, email addresses, etc.). Each entity type has its own recognition and validation rules, allowing the system to handle multiple text types independently and systematically, managing complexity through structured division of labor.
Solution Approach 2:
The patent implements a universal text recognition framework that can identify and process multiple types of text entities using a common architecture. The system uses a unified approach to detect, extract, and validate various text entities (phone numbers, URLs, emails, addresses) through shared preprocessing and postprocessing routines, achieving versatility without proportionally increasing complexity.
4Reliability
If error correction mechanisms are added to OCR, then recognition reliability improves, but processing time increases
Solution Approach 1:
The patent implements feedback mechanisms where the system validates recognized text entities against known patterns and formats (phone number formats, URL structures, email patterns). When recognition results are ambiguous or不符合 expected patterns, the system applies corrective algorithms and reprocesses the text, using feedback loops to improve reliability while managing processing time through selective correction.
Data Source
AI summary
Various embodiments enable the identification of semi-structured text entities in an imager. The identification of the text entities is a relatively simple problem when the text is stored in a computer and free of errors, but much more challenging if the source is the output of an optical character recognition (OCR) engine from a natural scene image. Accordingly, output from an OCR engine is analyzed to isolate a character string indicative of a text entity. Each character of the string is then assigned to a character class to produce a character class string and the text entity of the string is identified based in part on a pattern of the character class string.