Image Processing Using Visible and Invisible Light for Foggy Subjects

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing image processing techniques struggle to enhance visibility of subjects obscured by fog or haze, particularly when conventional subject detection methods fail to accurately identify main subjects in such conditions.

Innovation Solution

An image processing apparatus utilizing both visible and invisible light cameras, combined with deep learning models, performs gradation correction by integrating subject evaluation values from both light sources, adjusting weights based on fog and haze levels or distance, to accurately identify and enhance the main subject.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If correction processing is performed on the entire image using luminance histogram, then the luminance level of the image is corrected over all gradation ranges, but correction suitable for the main subject cannot be performed when the subject is covered in fog or haze

Engineering Contradiction:
Improveluminance correction accuracyVSAvoidsubject-specific correction capability
Core Design Contradiction:
Measurement precisionVSAdaptability or versatility

Solution Approach 1:

The image is divided into multiple regions including a sky region and a ground region. The sky region is further segmented to identify foggy areas. This segmentation allows different correction strategies to be applied to different regions, enabling subject-specific correction while maintaining overall image quality.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

Different luminance correction curves are generated and applied to different regions of the image. The sky region receives one correction curve optimized for fog removal, while the ground region receives another correction curve. This local quality approach ensures that each region is corrected according to its specific characteristics and fog conditions.

Inventive Principle:
Principle #3Local quality

2Measurement precision

If known subject detection techniques are used in combination with luminance correction, then detection of the main subject may be improved, but detection itself cannot be performed if the subject is particularly strongly covered in fog or haze

Engineering Contradiction:
Improvesubject detection accuracyVSAvoiddetection reliability in heavy fog
Core Design Contradiction:
Measurement precisionVSReliability

Solution Approach 1:

The system performs preliminary luminance correction on the sky region before subject detection. By first removing fog effects from the sky area using histogram analysis and generating an appropriate luminance correction curve, the system prepares the image data to enhance subsequent subject detection accuracy in foggy conditions.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The patent introduces an intermediate processing step that analyzes the luminance histogram of the sky region to generate a fog density map. This intermediate representation serves as a mediator that guides both the luminance correction process and the subject detection process, enabling reliable detection even when subjects are heavily obscured by fog.

Inventive Principle:
Principle #24Intermediary (Mediator)

Data Source

PatentEP4597417A1Image processing apparatus, image processing method, and computer-readable storage medium
Publication Date: 2025.08.06 CANON KK
  • EP4597417A1 patent drawingFigure 1
  • EP4597417A1 patent drawingFigure 2
  • EP4597417A1 patent drawingFigure 3A~3B

AI summary

An image processing apparatus comprises first detection means for detecting a subject from a first image obtained by image capturing using visible light, second detection means for detecting a subject from a second image obtained by image capturing using invisible light, and correction means for performing gradation correction on the first image based on a detection result of the subject by the first detection means and a detection result of the subject by the second detection means.