Perception-based color space for image segmentation application

After analyzing the deficiency of existing perception-based image processing problems, the most important properties of color spaces are identified without cross contamination between lightness, saturation, hue and with maximum perceptual uniformity. This paper using multi-grid optimization and standard color-difference formulas transformed images from an initial color space to the new color space. The new working color space performance ability is tested in image segmentation, using ISODATA algorithm to get the k-means clustering algorithms best cluster number and the initial cluster center. After comparing with the exiting color spaces, the experiment results showed color attributes of the new space free of cross contamination and with maximum perceptual uniformity which can improve the image segmentation algorithms´ accuracy.