Image Hue Determination Using Discrete Candidate Color Segmentation

The patent segments the continuous hue spectrum into discrete candidate colors (e.g., red, green, blue, yellow, cyan, magenta) with predetermined hue values. Instead of calculating a continuous average hue that may fall between color boundaries and cause distortion, the method divides pixels into groups based on their closest matching candidate color, then determines the image hue as one of these discrete candidate colors. This segmentation resolves the contradiction by sacrificing continuous precision for discrete accuracy that aligns with human color perception.

The patent inverts the traditional approach by not calculating an average hue value directly from all pixels. Instead, it assigns each pixel to a candidate color group and counts the weighted occurrences. The image hue is determined by the candidate color with the highest weighted count, rather than by averaging all hue values. This inversion of the calculation logic eliminates the distortion problem inherent in direct averaging while maintaining computational efficiency.

The patent segments the hue determination process into discrete candidate color categories. Each pixel is assigned to one of several candidate colors based on its hue value, and the image hue is determined by counting weighted occurrences in each category. This segmentation enables rapid determination by avoiding complex continuous calculations while accurately identifying the dominant color category, thus resolving the contradiction between speed and accuracy.

The patent introduces candidate colors as intermediary categories between individual pixel hues and the final image hue determination. Instead of directly averaging all pixel hues (which causes distortion), the method uses candidate colors as mediators to group and aggregate pixel information. This intermediary approach maintains calculation speed while eliminating distortion by leveraging the discrete nature of human color perception.