I understand that u performed some quality metric and u get negative value for Fused image. So negative value indicates there is possible amount of spatial and spectral distortion. It may be due to misregistration, Input band may not well correlated. And if you decomposed the MS image that also affect the value.  

Hello Salari, Please go through these points may helpful to you,

To maintain the spectral integrity, you would have to make sure that both the multispectral and pan images are in the same spectral space before you use them together in a pan sharpening. So, it sounds like you will need to correct your pan image to be in radiance or reflectance (whichever your corrected multispectral image is - probably reflectance). Or if your multispectral and pan images were collected by the same sensor, then you could do pan sharpening with the raw data, and then apply an atmospheric correction to the result of that.

please go through this  flow diagram

I hope that helps.

hello and many thanks for the answer!

Actually both pan image and multispectral images are from the same sensor (Landsat Oli) and I did atmospheric correction for both simultaneously through atmospheric correction wizard in Geomatica software in order to get reflectance images.

After that I tried to pan-sharpen multispectral reflectance images using Gram schmidt algorithm in Envi, but I got no satisfying result... there were negative values in all bands. More over true color composite image seems violet not natural.

Again I did pan-sharpening in Geomatica ( Geomatica uses its own algorithm for the pan-sharpening ). This time the colors are normal but there are still some negative values in all bands.I set the negative values to No-data. Is it rational?

Regards

One possible reason that the resulting image contains negative values is that the haze DN values used as input for the atmospheric correction are not valid (that is, they do not represent 'valid dark objects' within the image and/or the image area being used does not contain valid haze values).  If this is the case the image will be over corrected and not only will negative values be generated, but also the correction does not follow a realistic atmospheric scattering model (see my papers dealing with the 'Improved DOS' and 'COST' models for more details).

Also, depending on the procedure / method used for pan sharpening the spectral characteristics of some of the surface cover types can be changed and affected so that cover types that were separable in the original image are not in the resulting image.  For details on some of this look at the following 2 papers I published on pan sharping (I am sure there are other papers published by other authors as well):

Digital merging of Landsat TM and digitized NHAP data for 1:24 000-scale image mapping.

Photogrammetric Engineering and Remote Sensing, Vol. 52, No. 10, 1986, pp. 1637-1646.

------------------------------------------------------------

Comparison of Three Different Methods to Merge Multiresolution and Multispectral Data: Landsat TM and SPOT Panchromatic

Photogrammetric Engineering and Remote Sensing, Vol. 57, No. 3, March 1991, pp. 295-303

Good luck in your research.

Pat