Color Texture Analysis

As a measure for the correlation between two color planes the color covariance is defined by

is used to denote the color component of the pixel at position (x,y) of the image region I with |I| pixels. is the mean value and the standard deviation in the color plane i. The color covariance is calculated for a set of different displacement vectors in a two dimensional orthogonal image topology. Square matrices , , can be defined for a constant visibility distance D. In the centre of these covariance matrices the covariance measure of two color planes is located without a displacement.

It is , where 1 means high correlation, 0 no correlation and -1 anti-correlation. Inside the color planes i we have . A symmetrical characteristic is = . Therefore only half covariance matrices have to be calculated for all nine combinations of color planes explicitly: , , , , , , , , .