Signal reconstruction using discret wavelet

I am doing a discrete wavelet transform on a Lena.bmp image using the code

           R=(double(imread('10.bmp')));
LS = liftwave('haar','Int2Int');
[AR,HR,VR,DR] = lwt2(R,LS);


I get the Approximation part, Horizontal part, Vertical part, and the Diagonal part. Now I want to know which of these parts has more weightage in the image (I am sorry if I do not make myself clear with the term weightage). I think that the Approximation part is most important. Then after this, I do not know. For this what I did , I took the inverse of the wavelet transform and instead of the appropriate subband I just put an empty matrix in its place and calculate the PSNR between the reconstructed image and the original image. For instance if I wanted to know the weightage of the Horizontal subband I did the following

                RECONSTRUCTED=((ilwt2(AR,zeros(256,256),VR,DR,LS)));
psnr(uint8(R),uint8(VMCI))


and got the PSNR as 36.8702. Similarly for the Vertical subband I did

                       VMCI=((ilwt2(AR,HR,zeros(256,256),DR,LS)));
psnr(uint8(R),uint8(VMCI))


and got the PSNR as 33.5928 for the Diagonal subband I did

             VMCI=((ilwt2(AR,HR,VR,zeros(256,256),LS)));
psnr(uint8(R),uint8(VMCI))


and got the PSNR as 41.8740

Whati inferred from this was (even though it is justfor one image) that after the Approximation part, the Vertical detail subband is the most important followed by horizontal and then Diagonal subband. Is my inference correct or does this depend upon the signal/image? Can somebody help with this?