Image Processing Apparatus for Fog Removal via Region-Specific Parameters
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image processing techniques for removing fog and haze from surveillance camera images often apply uniform parameters, leading to unnatural images when processing subjects at different distances, and fail to enhance visibility of specific objects within thick fog areas.
Innovation Solution
An image processing apparatus that sets distinct parameters for fine particle removal based on the shooting scene and object visibility, using object detection to vary processing effects between regions, combining images processed with different parameters to enhance object visibility while minimizing unnatural effects.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If uniform parameters are used for fine particle removal processing across the entire image, then the processing is simple and fast, but the image appears unnatural when processing subjects at different distances
Solution Approach 1:
The image is divided into multiple regions based on distance from the camera (first region for close-distance subjects, second region for far-distance subjects). Different fine particle removal parameters are applied to each region, allowing optimized processing for each distance zone while maintaining overall processing efficiency.
Solution Approach 2:
Different parameter sets are assigned to different spatial regions of the image. The first parameter set is used for the first region (close-distance subjects) and the second parameter set is used for the second region (far-distance subjects), enabling locally optimized image quality without requiring full image complexity.
2Reliability
If fine particle removal processing is applied to the entire image, then visibility improvement is achieved, but objects in regions with already good visibility appear unnatural
Solution Approach 1:
The image is segmented into a first region where fine particle removal processing is applied and a second region where it is not applied (or applied with different parameters). This segmentation allows visibility improvement in foggy areas while preserving natural appearance in areas with already good visibility.
Solution Approach 2:
Fine particle removal processing is applied partially rather than uniformly across the entire image. The processing is selectively applied to regions where it is needed (first region) while avoiding or reducing application in regions where visibility is already adequate (second region), preventing over-processing and unnatural appearance.
3Reliability
If strong fine particle removal processing is applied to improve visibility of objects in thick fog, then visibility is improved, but the image becomes unnatural due to excessive processing
Solution Approach 1:
Different parameter sets with different processing strengths are applied to different regions. The first parameter set provides stronger processing for regions with thick fog (first region) to improve object visibility, while the second parameter set provides milder or no processing for regions with good visibility (second region) to maintain natural appearance.
Data Source
AI summary
This disclosure provides an image processing apparatus comprising a first setting unit which sets a first parameter for processing for removing an influence of a fine particle component based on image data; a first image processing unit which performs fine particle removal processing based on the first parameter; a second setting unit which sets a second parameter; a second image processing unit which performs fine particle removal processing based on the second parameter; a setting unit which sets a region for which the first image processing unit is to be used and a region for which the second image processing unit is to be used; and a generation unit which generates image data by applying a result from the first image processing unit and a result from the second image processing unit to the respective set regions.