Study on Correlation of Color Components Image in Different Color Spaces

Abstract

Color image is an information substance with color components, among which exists correlation. In order to investigate the internal relevance of information in color image and to find out the dependence between them, to research the correlation of the color component image is significant. In order to investigate the relationship between the color space and the correlation of the component images, color spaces RGB/ LCH/ LAB/ OHTA/ YCC are selected and the correlation coefficients and cross correlations of the component images are computed and analyzed on MATLAB platform. The Result shows, that the statistical correlation coefficients of component images under RGB color space are the highest, while in OHTA color space the lowest are showed. The correlation coefficients under LAB and LCH are relative lower. In the opposite color spaces, the correlation coefficients of two opposite color components images are higher than the coefficients between the lightness and one of the opposite color component images. For the cross correlation of color component images, it shows a weak negative exponent relationship between pixel distance and cross correlation. The average cross correlation of component images in LCH space is obvious lower than in other spaces, while the levels of cross correlation in other spaces are similar. The relationship between cross correlation and color characteristics of image in RGB color space is closely, while in OHTA space, the difference of cross correlations among component images are usually small. In LCH space, the difference of cross correlations among component images is obvious, the cross correlation among chroma and the other components (lightness and hue) are much lower.