Color Enhancement Method Using Luminance-Reflection Decomposition

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing image enhancement methods cause over-enhancement in regions with high luminance, leading to loss of detail, and introduce noise in regions with low luminance, due to the multiplication of color channels by a luminance enhancement ratio.

Innovation Solution

The method involves obtaining a luminance component, performing Gaussian filtering to get an illumination component, adjusting the reflection component, calculating a luminance gain, and applying enhancement processing to the image using this gain, while also decomposing the image into RGB components, performing wavelet low-pass filtering, and linear stretching to maintain color fidelity and reduce noise.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Manufacturing precision

If the three color channels of the original image are multiplied by the luminance enhancement ratio to enhance the image, then the contrast of the image is improved, but regions with high luminance become too bright causing over-enhancement and loss of detail, while regions with low luminance introduce noise

Engineering Contradiction:
Improveimage enhancement qualityVSAvoiddetail preservation and noise control
Core Design Contradiction:
Manufacturing precisionVSReliability

Solution Approach 1:

The image processing is segmented into multiple independent pathways: one for luminance enhancement (affecting brightness and contrast) and another for color channel processing (preserving color information). The luminance component is enhanced separately from the color components, allowing selective enhancement without introducing noise to color channels or causing over-enhancement in high-luminance regions.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The luminance component is extracted from the original image using multi-scale Gaussian filtering to obtain the illumination map. This extracted luminance component is then enhanced independently through histogram equalization or other enhancement techniques, while the original color channels are preserved without direct multiplication, avoiding the introduction of noise and over-enhancement artifacts.

Inventive Principle:
Principle #2Taking out (Extraction)

2Manufacturing precision

If multi-scale Gaussian filtering is performed on the luminance component to obtain the illumination component, then the reflection component is isolated, but the processing complexity increases

Engineering Contradiction:
Improvereflection component isolationVSAvoidfiltering processing complexity
Core Design Contradiction:
Manufacturing precisionVSDevice complexity

Solution Approach 1:

The filtering process uses dynamic multi-scale Gaussian kernels with varying standard deviations to adapt to different local characteristics of the luminance component. This allows the algorithm to effectively separate illumination and reflection components across different spatial frequencies without requiring overly complex fixed-structure filters.

Inventive Principle:
Principle #15Dynamics

Solution Approach 2:

The problem is extended from single-scale to multi-scale analysis by applying Gaussian filtering at multiple standard deviations. This dimensional extension in the scale space allows better separation of illumination and reflection components by capturing information at different spatial resolutions, improving accuracy while managing complexity through hierarchical processing.

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

Data Source

PatentEP2806395B1Color enhancement method and device
Publication Date: 2020.01.01 HUAWEI TECH CO LTD
  • EP2806395B1 patent drawingFigure 1
  • EP2806395B1 patent drawingFigure 2
  • EP2806395B1 patent drawingFigure 3

AI summary

Embodiments of the present invention disclose a method and an apparatus for enhancing color, and relate to the image processing field, so as to solve a case of over enhancement on a region having high luminance in an original image which is caused after enhancement processing is performed, and eliminate a large amount of noise that occurs on a region having low luminance in the original image after the enhancement processing is performed. A specific solution is: obtaining a luminance component of a currently processed image; performing Gaussian filtering processing on the luminance component to obtain an illumination component of the currently processed image; obtaining a first reflection component of the currently processed image according to the luminance component and the illumination component; performing enhancement adjustment on the first reflection component to obtain a second reflection component; obtaining a luminance gain according to the luminance component and the second reflection component; and performing enhancement processing on the currently processed image according to the luminance gain, to obtain a first enhanced image. The present invention is applicable to a process of color enhancement.