Mobile OCR Text Segmentation and Alignment
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current systems face challenges in seamlessly searching electronic documents based on paper documents, requiring manual review and lacking efficient optical character recognition (OCR) capabilities on mobile devices due to inferior camera performance, skewed or rotated images, and high processing power requirements.
Innovation Solution
Implementing a system that captures images of a paper document in portions, performs OCR on each portion, joins the text using natural language processing, and uses motion data from the mobile device to determine the order and alignment of the text, thereby reducing processing time and improving accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If OCR is performed on the entire document at once, then complete text recognition is achieved, but processing time and power consumption increase significantly
Solution Approach 1:
The patent divides the document into multiple portions or regions, capturing and processing images of each portion separately. This segmentation allows the OCR system to handle smaller image segments sequentially or in parallel, significantly reducing the processing time and computational load compared to processing the entire document at once, while still achieving complete text recognition by combining results from all portions.
2Measurement precision
If high-resolution images are captured for accurate OCR, then text recognition accuracy improves, but camera performance requirements and processing power increase
Solution Approach 1:
By segmenting the document into smaller portions, the system can capture images at appropriate resolution for each segment without requiring the entire document to be captured at maximum resolution simultaneously. This reduces the overall processing power requirements while maintaining text recognition accuracy for each segment.
Solution Approach 2:
The patent extracts only the necessary portions of the document for processing at any given time, rather than processing the entire document. This extraction approach allows the system to use moderate camera resolution and processing power for each extracted portion, reducing the cumulative power requirements while achieving complete text recognition.
3Measurement precision
If manual review is performed to search electronic documents based on paper documents, then search accuracy is maintained, but time consumption and operational complexity increase
Solution Approach 1:
The system performs automated OCR processing on captured document portions, extracting text without requiring manual review. This self-service approach maintains search accuracy through automated text recognition while dramatically improving productivity by eliminating the time-consuming manual review process.
Solution Approach 2:
The patent replaces the mechanical manual review process with an automated optical character recognition system. This substitution maintains search accuracy through sophisticated OCR algorithms while improving productivity by automating the text extraction process, eliminating the need for human operators to manually review and transcribe document text.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
The present invention provides a method for document searching. The method for document searching comprises: capturing a first image of a first portion of a document at a device; determining a first text associated with the first image, wherein the first text is determined by performing optical character recognition (OCR) on the first image; receiving a first result from a first search of a set of documents that was performed based on one or more search terms determined based on the first text associated with the first image; and presenting the first result on the device. Additionally, the present invention provides a method for performing optical character recognition (OCR) on a device.