Document Image Shade Removal via Localized Correction

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing methods for removing shade from captured images, particularly in mobile terminals, often result in overcorrection, leading to improper brightness adjustments of objects and backgrounds, reducing the quality of printed images.

Innovation Solution

A method and apparatus for image processing that estimates shade components based on image capture environment conditions, using a reduced image to calculate correction rates and adjust pixel values to prevent overcorrection, ensuring accurate removal of shade while preserving object colors.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If shade component removal correction is performed on all regions including objects, then shade in the background is removed, but the color of objects turns bright due to excessive correction

Engineering Contradiction:
Improveshade removal accuracyVSAvoidobject color accuracy
Core Design Contradiction:
Measurement precisionVSManufacturing precision

Solution Approach 1:

The patent applies different correction strategies to different regions of the image. Character regions are detected and corrected with a small correction amount to prevent overcorrection, while background regions are corrected with a larger correction amount to effectively remove shade. This local differentiation resolves the contradiction by allowing aggressive shade removal in backgrounds while preserving object colors.

Inventive Principle:
Principle #3Local quality

Solution Approach 2:

The image is segmented into character regions and background regions based on detected features. By dividing the image processing task into separate handling for text areas and background areas, the patent can apply appropriate correction amounts to each segment, preventing overcorrection of objects while effectively removing background shade.

Inventive Principle:
Principle #1Segmentation

2Productivity

If correction amount is increased to remove shade effectively, then shade removal performance improves, but objects are overcorrected and turn bright

Engineering Contradiction:
Improveshade removal efficiencyVSAvoidcolor accuracy
Core Design Contradiction:
ProductivityVSManufacturing precision

Solution Approach 1:

The patent implements spatially varying correction amounts where the correction strength depends on the local region type. Background regions receive strong correction for efficient shade removal, while object/character regions receive minimal correction to maintain color accuracy. This resolves the productivity-precision contradiction through localized parameter adjustment.

Inventive Principle:
Principle #3Local quality

Solution Approach 2:

The patent applies partial correction to character regions (small correction amount) and excessive correction to background regions (large correction amount). This selective application of correction intensity allows the system to achieve both goals: effective shade removal in backgrounds and color preservation in objects.

Inventive Principle:
Principle #16Partial or excessive action

Data Source

PatentEP3119074B1Information processing apparatus, method for processing information, and computer program
Publication Date: 2021.03.24 CANON KK
  • EP3119074B1 patent drawingFigure 1
  • EP3119074B1 patent drawingFigure 2A~2B
  • EP3119074B1 patent drawingFigure 3

AI summary

An information processing method includes calculating at least one difference value between a pixel value of a target pixel in an input image obtained by capturing a document image and a pixel value of a corresponding pixel in a shade component image at coordinates identical to those of the target pixel. The shade component image represents shade components of the input image. The shade components depend on an image capture environment. The method further includes correcting the pixel value of the target pixel in the input image based on a correction rate to correct a shade component in the target pixel. The correction rate depends on at least one difference value associated with the target pixel in the input image.