Chrominance similarity inof images in matlab -

i doing project on image quality assessment , implementing following paper

a new method color image quality assessment - niveditta thakur , swapna devi, international journal of computer applications, february 2011

i have implemented of code struggling understand how calculate color similarity. converted rgb image hsi , , want extract chrominance information in single plane , similarity measure.

here did:

hsi=rgb2hsi(i); h = hsi(:,:,1); % seperate saturation & intensity give hue of image s = hsi(:,:,2); % seperate hue & intensity give saturation of image = hsi(:,:,3); % seperate hue & saturation give intensity/grayscale image h_n=cat(3,h,s,x_z); g_n=rgb2gray(h_n); k=1; x=1:factor:s1-2     y=1:factor:s2-2          i=0:factor-1             j=0:factor-1                 n1((i+1),(j+1),k)=g_n((x+i),(y+j),1);             end         end         k=k+1;     end end  p=1:3072     mh1(p)=mean2(n1(:,:,p)); end 

i ran on both distorted , test image. similarity measure computed follows:

hue=[((2*(mh1*mh2)))]/[(((mh1)^2+(mh2)^2))]; 

but still don't have desired results.

can help?