Text Recognition Color Inversion for Stable Mobile Imaging

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Conventional camera-based text recognition methods face challenges in maintaining stable recognition capability due to diverse text inputs and unpredictable environments, particularly with inverted text images where the text color is brighter than the background, leading to difficulties in proper pre-processing and post-processing.

Innovation Solution

A text recognition apparatus with a text region detection unit that discriminates text and background colors, unifies non-text background regions into the text-background color, and applies inversion-processing when necessary to improve text recognition accuracy, ensuring correct recognition even in varied environments.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If conventional text recognition methods are used with camera input, then text acquisition is easy and real-time recognition is possible, but recognition stability deteriorates due to diverse text types and unpredictable environments

Engineering Contradiction:
Improvereal-time text recognition speedVSAvoidrecognition stability
Core Design Contradiction:
ProductivityVSReliability

Solution Approach 1:

The patent applies preliminary action by performing inversion processing before text recognition. The system determines whether text images are inverted (text brighter than background) and applies inversion processing to convert them to normal form before recognition, thereby improving recognition stability in advance without affecting real-time performance

Inventive Principle:
Principle #10Preliminary action

2Measurement precision

If conventional pre-processing is applied, then standard text images are processed correctly, but inverted text images (text brighter than background) cannot be properly recognized

Engineering Contradiction:
Improvetext recognition accuracyVSAvoidhandling of diverse text types
Core Design Contradiction:
Measurement precisionVSAdaptability or versatility

Solution Approach 1:

The patent applies dynamics by making the pre-processing approach adaptive rather than fixed. The system dynamically determines whether text images are inverted based on comparing text brightness with background brightness, and selectively applies inversion processing only when needed. This dynamic adaptation allows the system to handle both normal and inverted text images effectively

Inventive Principle:
Principle #15Dynamics

Solution Approach 2:

The patent directly applies inversion by detecting inverted text images where text is brighter than background and applying inversion processing to swap the brightness relationship. This converts inverted text images into normal form that can be properly recognized by standard text recognition algorithms

Inventive Principle:
Principle #13The other way round (Inversion)

3Measurement precision

If no background unification is performed, then processing is faster, but text region extraction accuracy deteriorates due to non-uniform backgrounds

Engineering Contradiction:
Improvetext region extraction accuracyVSAvoidprocessing time
Core Design Contradiction:
Measurement precisionVSLoss of time

Solution Approach 1:

The patent applies local quality by performing background unification only in the text-background regions surrounding the text, rather than processing the entire image. This localized approach maintains text region extraction accuracy by ensuring uniform backgrounds around text while minimizing unnecessary processing time on other image regions

Inventive Principle:
Principle #3Local quality

Data Source

PatentEP2221746B1Apparatus and method for improving text recognition capability
Publication Date: 2019.03.06 SAMSUNG ELECTRONICS CO LTD
  • EP2221746B1 patent drawingFigure 1
  • EP2221746B1 patent drawingFigure 2
  • EP2221746B1 patent drawingFigure 3

AI summary

Disclosed are an apparatus and a method for text recognition capability using a camera provided in a mobile communication terminal. Image pre-processing discriminates a text color and a text-background color in an input image, and unifies regions except the text into the text-background color, so that a text region and a background region surrounding the text region can be precisely separated. The image pre-processing method is adaptive to a photographing environment, whereby stable text recognition capability can be expected even if the photographing environment is variously changed.