Red-Eye Detection Using Multi-Parameter Color Evaluation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional automatic red-eye correction methods in digital cameras and printers face challenges in accurately detecting red-eye regions, especially for individuals with dark pigments, due to the reliance on skin color detection and saturation differences, which are not effective and require significant computational resources.
Innovation Solution
An image processing apparatus and method that calculates an evaluation amount for each pixel based on predetermined color components, extracts candidate pixels, and performs determinations on characteristic amounts such as hue, saturation, and edge intensity to accurately detect red-eye regions, regardless of pigment type.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If conventional skin color detection and saturation difference methods are used for red-eye detection, then the detection process is simple, but the detection accuracy deteriorates for individuals with dark pigments
Solution Approach 1:
The patent changes the detection parameters from relying solely on saturation differences to using a comprehensive evaluation that includes hue angle, saturation, and luminance. By introducing the hue angle parameter and evaluating multiple color space components simultaneously, the system achieves accurate red-eye detection across different skin tones without increasing operational complexity
Solution Approach 2:
The patent introduces an intermediary evaluation mechanism that calculates an evaluation amount for each pixel based on multiple color components (hue angle, saturation, luminance) before making the red-eye determination. This intermediary step synthesizes multiple parameters into a unified assessment, enabling accurate detection while maintaining simple operation
2Device complexity
If conventional red-eye detection methods are used, then the device complexity is low, but the computational resources required increase significantly for accurate detection
Solution Approach 1:
The patent segments the detection process into distinct functional steps: calculating color space conversion for each pixel, determining hue angle and saturation values, evaluating multiple color components, and making the final red-eye determination. This segmentation allows efficient processing by handling each step with appropriate computational complexity, avoiding unnecessary resource consumption while maintaining accuracy
3Manufacturing precision
If manual or semiautomatic correction methods are used, then the correction precision is high, but the ease of operation deteriorates due to manual designation requirements
Solution Approach 1:
The patent implements automatic red-eye detection and correction without requiring manual user intervention. The system autonomously detects red-eye regions using the multi-parameter evaluation method and executes correction processing automatically, combining the precision of manual correction with the convenience of automated operation. The detection unit and correction unit work together to provide self-service functionality
Data Source
AI summary
An image processing apparatus configured to detect an image region indicating a poor color tone of eyes from candidate regions includes: a first determination unit configured to perform a determination relating to an evaluation amount with respect to the poor color tone based on a predetermined color component in a target candidate region, and the evaluation amount in a peripheral region of the candidate region, a second determination unit configured to update the candidate region based on the first determination result, and to perform a determination relating to the evaluation amount or a predetermined color component with reference to pixels in an updated second candidate region, and a third determination unit configured to update the second candidate region based on the second determination result, and to perform a determination relating to a characteristic amount of a peripheral region with reference to pixels in a peripheral region of an updated third candidate region. A calculation amount of the third determination is greater than a calculation amount of the first or second determination.