I'm not sure if you did this step or not. You need to split your color image into its 3 bands: Red, Green, & Blue and match each band seperatly as a standalone gray level image. Then, you avergae the correlation factor for the 3 bands. Also you may need to convert the color image into a gray image and then carry the correlation on the gray image.
Yes, I agree with the second reccomendation of Aparna: convert the image to grayscale to use in 'normxcorr2'. Try the example on the Mathworks website: