X
PrevPrev Go to previous topic
NextNext Go to next topic
Last Post 11 Oct 2007 01:01 AM by  anon
Problem with PC Sharpening of ASTER MX with RadarSAT-1
 1 Replies
Sort:
You are not authorized to post a reply.
Author Messages

anon



New Member


Posts:
New Member


--
11 Oct 2007 01:01 AM
    Dear Fellow remote sensors, Running ENVI 4.3 on WinXP service pack 2 I am currently trying to fuse a RadarSAT-1 image (res 6.25m) to an ASTER MX (res 15m) data set. Essentailly the SAR image is acting like a pseudo panchromatic band where it is used to inject texture and improve the spatial resolution of the ASTER data. I have used the brovey and HSV approaches with no problems, however, when I attempt to use the PC sharpening approach I run into problems. Following the completion of the module the resulting imagery looks okay but when I look at the individual bands it is clear that there are problems with the fusion. Basically, SAR texture has been added to the NIR band but the green and red bands have merely been replaced with the SAR image resulting in severe spectral distortion. I would like to know if anyone else has had this problem, I have repeated the exercise using the resolution merge application in ERDAS with similar results. The literature indicates that PCA is a fairly good fusion procedure and this is unexpected. Does anyone have any suggestions? (both images are integer 0 - 255) Regards, Wesley

    MariM



    Veteran Member


    Posts:2396
    Veteran Member


    --
    11 Oct 2007 01:01 AM
    I am not sure what is causing the color distortion but I am fairly certain it is something in the data. I have used PC Sharpening (with traditional pan bands) and it gives very good results. I am not clear about your description of the result - the SAR being added to the NIR but green and red being 'replaced' by the SAR image. This is not how PC Sharpening works. There is the forward PC transform, the first PC band is replaced with the high resolution band (SAR in your case), and the PC + SAR bands are inversed transformed using the PCA statistics. The resulting sharpened image cannot be a simple 'replacement' of the G and R bands with the SAR band. It also sounds like the combined bands of the result look ok but the single bands do not? I just am not very clear about what is going on and it might be easier if I could see the data. If you like, contact Technical Support for more assistance.
    You are not authorized to post a reply.