Dynamic Image Color Correction via Sub-Region Noise Analysis

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing image processing systems struggle to accurately identify the color of objects, particularly vehicles, due to factors like light conditions and obstructed license plates, leading to poor vehicle color identification capabilities.

Innovation Solution

A system and method for image processing that adjusts a modification matrix based on noise values and channel correlation coefficients of RGB channels to correct image colors, ensuring accurate color identification without requiring repeated calibration of the acquisition environment.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Device complexity

If a fixed color correction matrix is used for image processing, then the device complexity is reduced, but the color accuracy deteriorates under varying light conditions and noise levels

Engineering Contradiction:
Improvecolor correction system complexityVSAvoidcolor identification accuracy
Core Design Contradiction:
Device complexityVSMeasurement precision

Solution Approach 1:

The patent implements dynamic color correction by dividing the image into multiple sub-images and calculating separate correction matrices for each sub-image based on local noise characteristics. This transforms the static color correction approach into a dynamic one that adapts to varying conditions across different regions of the image, thereby maintaining color accuracy without requiring overly complex global calibration systems.

Inventive Principle:
Principle #15Dynamics

Solution Approach 2:

The patent applies local quality by processing different regions of the image with different correction parameters. Each sub-image receives a customized correction matrix calculated from its specific noise levels and channel correlation coefficients, rather than applying a uniform correction across the entire image. This localizes the correction quality to match local conditions, improving overall color accuracy.

Inventive Principle:
Principle #3Local quality

2Measurement precision

If repeated calibration of the acquisition environment is performed to improve color accuracy, then the color identification accuracy improves, but the loss of time increases

Engineering Contradiction:
Improvecolor identification accuracyVSAvoidcalibration time
Core Design Contradiction:
Measurement precisionVSLoss of time

Solution Approach 1:

The patent performs preliminary noise estimation and channel correlation coefficient calculation for each sub-image before applying color correction. By pre-calculating the necessary parameters (noise levels, correlation coefficients) and organizing them into correction matrices in advance, the system eliminates the need for repeated environmental calibration during actual image processing, saving time while maintaining accuracy.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The system performs self-calibration by automatically estimating noise characteristics and calculating correction matrices from the image data itself, without requiring external calibration targets or manual intervention. The algorithm uses the image's own statistical properties (noise values, channel correlations) to generate appropriate correction parameters, making the calibration process autonomous and time-efficient.

Inventive Principle:
Principle #25Self-service

3Measurement precision

If spatially varying color correction is applied to improve color accuracy in different image regions, then the color identification accuracy improves, but the device complexity increases

Engineering Contradiction:
Improvecolor correction accuracyVSAvoidimage processing system complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent segments the image into multiple sub-images and processes each segment independently with its own correction matrix. This segmentation approach enables spatially varying color correction by allowing different correction parameters for different regions, while keeping the computational complexity manageable through localized processing rather than requiring complex global optimization algorithms.

Inventive Principle:
Principle #1Segmentation

Data Source

PatentEP3625761B1Systems and methods for image processing
Publication Date: 2023.08.02 ZHEJIANG DAHUA TECH CO LTD
  • EP3625761B1 patent drawingFigure 1
  • EP3625761B1 patent drawingFigure 2
  • EP3625761B1 patent drawingFigure 3

AI summary

A method for identifying a color of an object may include obtaining a target image including the object and identifying in the target image a region of interest (ROI) including the object. The method may also include identifying at least one sub-region of the object from the ROI of the object and determining at least one first characteristic vector respectively corresponding to the at least one sub-region. The method may further include determining a second characteristic vector of the ROI of the object and identifying, based on the at least one first characteristic vector and the second characteristic vector, the color of the object using a color identification model. The method may further include correcting the color of the target image.