OCR Character Positioning via Line Word Indices

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing optical character recognition (OCR) techniques are cumbersome and time-consuming due to the reliance on pixel coordinates for character positioning, requiring extensive processing and image analysis.

Innovation Solution

The method involves extracting character images along with position indices, including line, word, and character numbers, using an OCR algorithm to generate text by positioning characters based on these indices, reducing the need for complex pixel coordinate calculations.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If pixel coordinates are used for character positioning in OCR, then character position accuracy is maintained, but processing time and computational complexity increase significantly

Engineering Contradiction:
Improvecharacter position accuracyVSAvoidprocessing time
Core Design Contradiction:
Measurement precisionVSLoss of time

Solution Approach 1:

The patent segments the character positioning process into two distinct phases: (1) image analysis phase where character bounding boxes are identified and stored with their pixel coordinates, and (2) text generation phase where pre-stored position data is retrieved and applied. This segmentation allows computationally intensive image analysis to be performed once, while subsequent text generation operations use pre-processed position data, significantly reducing processing time without sacrificing position accuracy.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent performs preliminary action by pre-calculating and storing character position information (bounding boxes, line numbers, word positions) during the image analysis phase before actual text generation occurs. This pre-processing creates a position lookup table that can be quickly referenced during text generation, eliminating the need for repeated complex coordinate calculations and thereby reducing processing time while maintaining accuracy.

Inventive Principle:
Principle #10Preliminary action

2Measurement precision

If extensive image analysis is performed for character positioning, then positioning accuracy is improved, but device complexity and processing requirements increase

Engineering Contradiction:
Improvecharacter positioning accuracyVSAvoidprocessing complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent divides the complex image analysis task into manageable components: character detection, bounding box calculation, line identification, and word positioning. Each component processes specific aspects of character location independently, allowing the system to handle complexity in modular fashion rather than requiring a single complex positioning algorithm, thereby reducing overall device complexity while maintaining positioning accuracy.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent introduces an intermediary position data structure (bounding box with metadata including line number, word position, and pixel coordinates) that serves as a bridge between image analysis and text generation. This intermediary representation simplifies the interface between complex image processing and straightforward text output, reducing the complexity of coordinating these operations while preserving positioning precision.

Inventive Principle:
Principle #24Intermediary (Mediator)

Data Source

PatentUS10482344B2System and method for performing optical character recognition
Publication Date: 2019.11.19 WIPRO LTD
  • US10482344B2 patent drawing
  • US10482344B2 patent drawing
  • US10482344B2 patent drawing

AI summary

Disclosed herein is a system and method for performing OCR. The method includes receiving a textual image, and extracting one or more character images along with one or more corresponding position indices from the textual image. A position index of a character image may include a line number, a word number, and a character number of the character image within the textual image. The method further includes determining one or more characters corresponding to the one or more character images using an OCR algorithm, and generating a text by positioning the one or more characters based on the one or more position indices of the one or more corresponding character images.