Image Processing Apparatus for Low-Resolution Character Recognition

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current OCR processes for character images captured by cameras face significant challenges with low-resolution images, where recognition rates are severely reduced and vary greatly due to photography conditions, leading to difficulties in determining final characters, especially when high-quality images cannot be obtained.

Innovation Solution

An image processing apparatus and method that extracts and compares character strings from multiple images captured from different angles and positions, calculating similarity degrees and using weighting factors based on positional and pixel information to enhance recognition precision, allowing for accurate character string recognition even with low-resolution images.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If OCR process is executed on low-resolution character images, then more character images can be processed, but recognition rate extremely lowers

Engineering Contradiction:
Improvenumber of character images processedVSAvoidcharacter recognition rate
Core Design Contradiction:
ProductivityVSMeasurement precision

Solution Approach 1:

The patent combines multiple low-resolution character images showing the same document into a single high-quality image through super-resolution processing. By merging redundant information from multiple images, the system reconstructs a high-resolution image that maintains accurate character recognition while enabling processing of originally low-resolution inputs.

Inventive Principle:
Principle #5Merging (Combining)

Solution Approach 2:

The patent transitions from processing individual low-resolution 2D images to utilizing multiple 2D images in a temporal dimension (captured at different times) to create a high-resolution output. This multi-dimensional approach allows the system to overcome resolution limitations by leveraging information across multiple image captures.

Inventive Principle:
Principle #17Another dimension (Dimensionality change)

2Measurement precision

If OCR process is executed on only high-quality character images, then recognition rate is maintained, but all character images are excluded when high-quality images cannot be obtained

Engineering Contradiction:
Improvecharacter recognition rateVSAvoidability to process various photography conditions
Core Design Contradiction:
Measurement precisionVSAdaptability or versatility

Solution Approach 1:

The patent converts the harmful effect of multiple photographs taken under varying conditions into a beneficial resource. By using these previously discarded low-quality images as input for super-resolution processing, the system transforms photography variability from a problem into a solution, enabling high recognition rates even when individual images are of poor quality.

Inventive Principle:
Principle #22Blessing in disguise (Convert harm into benefit)

3Quantity of substance

If OCR process is executed on plurality of character images photographed at different timings, then more data is available, but determination of final characters is difficult when recognition results divide

Engineering Contradiction:
Improvenumber of recognition resultsVSAvoiddifficulty of determining final characters
Core Design Contradiction:
Quantity of substanceVSDevice complexity

Solution Approach 1:

The patent performs super-resolution processing and image merging before the OCR recognition step. By pre-processing multiple images into a single high-quality image, the system eliminates the need to handle multiple separate recognition results, simplifying the determination process while maintaining high accuracy.

Inventive Principle:
Principle #10Preliminary action

Data Source

PatentEP3079100B1Image processing apparatus, image processing method and computer readable storage medium
Publication Date: 2019.01.30 TOSHIBA TEC KK
  • EP3079100B1 patent drawingFigure 1~2
  • EP3079100B1 patent drawingFigure 3~4
  • EP3079100B1 patent drawingFigure 5

AI summary

According to one embodiment, an image processing apparatus includes a calculation unit and a recognition unit. The calculation unit is configured to calculate a first similarity degree group which is composed of similarity degrees between respective characters constituting a first character string appearing on a first image and respective candidate characters in a candidate character group, and to calculate a second similarity degree group which is composed of similarity degrees between respective characters constituting a second character string appearing on a second image and the respective candidate characters in the candidate character group.