Preferred Color Detection Using Hue Reference and Best Linear Estimation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for detecting preferred-color areas in image display devices require significant computational resources and are not accurate for colors like green and blue, as they are not defined by elliptical or rectangular shapes, and statistical methods also consume excessive computations.
Innovation Solution
A method using hue reference values and best linear estimation (BLE) to detect preferred-color pixels by converting RGB data into Lab color coordinates, calculating hue and chroma values, and performing a matrix product with weight vectors to determine preferred-color discrimination constants, reducing computational load while improving accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If elliptic or rectangular color area models are used for detection, then detection accuracy for certain colors (skin color) is improved, but computational load increases significantly due to computing elliptic or linear equations for all pixels
Solution Approach 1:
The patent transforms the detection approach by changing the parameter space from spatial coordinates (x, y) in elliptic/rectangular models to color space parameters (hue, saturation, value). This allows detection based on color characteristics rather than geometric equations, significantly reducing computational load while maintaining accuracy for skin colors and other preferred colors.
Solution Approach 2:
The patent segments the detection process into two stages: first detecting candidate pixels using hue value comparison with reference ranges, then performing BLE operation only on these candidates. This segmentation reduces the number of pixels requiring intensive computation from all image pixels to only those within preferred color hue ranges.
2Measurement precision
If statistical methods are used for color detection, then comprehensive color analysis is achieved, but computational load increases due to probability computation and threshold comparison for each pixel
Solution Approach 1:
The patent extracts only the essential color parameters (hue, saturation, value) needed for preferred color detection from the full color information. By focusing on these extracted parameters and using reference value comparisons with BLE operation, it achieves effective color analysis without the excessive computational burden of complete statistical probability computations for all pixels.
3Power
If hue and saturation ranges are used for detection, then computational load is reduced, but detection accuracy decreases as non-preferred colors are included in the detected area
Solution Approach 1:
The patent introduces a dynamic two-stage detection process that adapts to different color types. For skin colors, it applies both hue range filtering and BLE operation with skin color reference values. For other preferred colors (green, blue), it uses hue range filtering combined with saturation and value threshold adjustments. This dynamic approach maintains low computational load while improving accuracy by tailoring the detection criteria to each color type.
4Measurement precision
If elliptic model is used for skin color detection, then detection accuracy for skin color is improved, but the model becomes unsuitable for green or blue colors which are distributed over wide areas
Solution Approach 1:
The patent creates a universal detection framework based on color space parameters (hue, saturation, value) that can detect multiple types of preferred colors (skin color, green, blue, etc.) using the same fundamental approach. By using BLE operation with color-specific reference values and threshold adjustments, the system achieves both high accuracy for skin color and versatility for detecting other preferred colors distributed over wide areas.
Data Source
AI summary
Disclosed herein are a method and apparatus for detecting a preferred color, which is capable of accurately detecting a preferred-color area with a small amount of computations, and a liquid crystal display device using the same. The method for detecting a preferred color includes comparing a hue value of an input pixel with hue reference values and detecting a first preferred-color pixel, and performing a best linear estimation (BLE) operation with respect to the first preferred-color pixel and detecting a second preferred-color pixel.