Image Correction Apparatus Edge Preservation via Reduced Image Smoothing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image correction techniques, particularly those based on the Retinex theory, face challenges in reducing calculation complexity and artifact suppression, especially when dealing with high-definition images and abrupt changes in illumination, which can lead to inaccuracies in contrast enhancement.
Innovation Solution
An image correction apparatus that generates a reduced image, calculates a reflectance component, and determines a correction factor based on a histogram excluding edge pixels, thereby reducing calculation complexity and suppressing artifacts by preserving edge information during the image enlargement process.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If smoothing process is performed on the original image to estimate illuminance distribution, then the illuminance distribution can be obtained, but edge portions are lost and artifacts occur in the corrected image
Solution Approach 1:
The image processing is divided into two independent stages: (1) smoothing the reduced image to obtain illuminance distribution, and (2) applying correction to the original image using the obtained illuminance distribution. This segmentation allows the smoothing operation to be performed on a downsampled version, avoiding edge loss in the full-resolution image while still providing accurate illuminance estimation for contrast correction.
Solution Approach 2:
The patent introduces a resolution dimension by working with both a reduced image (lower resolution) and the original image (full resolution). The smoothing operation is performed in the reduced image domain, and the results are transferred back to the original image domain for correction, effectively separating the smoothing function from the edge preservation requirement.
2Measurement precision
If smoothing is performed on the original high-definition image, then accurate illuminance distribution is obtained, but the calculation amount becomes extremely large
Solution Approach 1:
Instead of performing computationally intensive smoothing operations on the full-resolution original image, the patent creates a reduced copy (downsampled version) of the image for the smoothing operation. This copy contains sufficient information for illuminance estimation but requires significantly fewer computational resources. The illuminance distribution obtained from the reduced copy is then applied to correct the original high-definition image.
3Loss of information
If edge pixels are included in histogram calculation, then the histogram represents the entire image, but the correction factor becomes inaccurate due to abrupt illuminance changes at edges
Solution Approach 1:
The patent extracts and excludes edge pixels from the histogram calculation process. By identifying pixels located at edges (where abrupt illuminance changes occur) and removing them from the statistical analysis, the histogram accurately represents only the regions where reflectance information is reliable. This leads to more accurate correction factors that are not skewed by edge artifacts.
Data Source
AI summary
An image correction apparatus generates a reduced smoothed image by smoothing a reduced image created from an input image; detects at least one edge pixel on the reduced image or the reduced smoothed image; calculates, for each pixel of the reduced image, a reflectance component of an object; calculates a correction factor having a value based on a reflectance component in accordance with a distribution of reflectance components other than the reflectance component of the edge pixel; generates an enlarged smoothed image by enlarging the reduced smoothed image; and generates a corrected image by calculating a luminance value of each pixel of the corrected image based on the correction factor and the reflectance component based on a ratio of a luminance value of a corresponding pixel of the input image to a luminance value of a corresponding pixel of the enlarged smoothed image.