X
PrevPrev Go to previous topic
NextNext Go to next topic
Last Post 25 Oct 2013 01:06 PM by  anon
Degree missing in GCP in Classic 5.0
 1 Replies
Sort:
You are not authorized to post a reply.
Author Messages

anon



New Member


Posts:
New Member


--
25 Oct 2013 01:06 PM
    We're investigating a classification from MODIS and using TM as ground truth. This involves warping a TM raster in UTM to an expanded MODIS-derived raster in its native satellite projection. This would be trivial in the Sat-UTM direction, but UTM->Sat must be done by hand, picking GCPs and warping with a degree-2 NN polynomial. There are certain functionalities I need to find equivalents to: a) display UTM grid over TM scene in half-degree increments b) construct similar vector overlay for MODIS using LAT and LON rasters by contouring c) build GCP between TM & MODIS, selecting from matching degree intersections in both images [There are not enough identifiable landmarks in scenes, so I use LL degrees.] d) minimize RMS error in table for degree 2 Poly e) warp TM to MODIS using 2nd degree polynomial Nearest Neighbor. Eventually, I'll have to use the new interface, but for now I find most of these in Classic mode. Except the GCP selector only displays errors for 1st degree -- no switch to bump up to deg 2 or 3 after enough points are accumulated. I find that twenty GCPs are sufficient to warp at deg 2. Note that the warp program itself allowed RST/Polynomial/Triang and NN/Bil/CC methods, but the RMS error display seems to be stuck on RST and Degree 1. The new Image-to-Image registration interface has the same lack as the Classic. I'm sure the programmers will get around to including it, if enough people complain. Let me be the first. Tim Szeliga National Operational Hydrologic Remote Sensing Center (NOHRSC) National Weather Service, NOAA PS: Looking at the "How to create ROIs in ENVI 5" page. The standard Land Use / Land Cover schema based on the USGS Anderson model has between nine and ninety-two categories. I usually use twenty. Does each category have to be a separate Shapefile?

    MariM



    Veteran Member


    Posts:2396
    Veteran Member


    --
    29 Oct 2013 07:04 AM
    Hi Tim, I am not sure I am seeing what you describe. I am working with a dataset and around 20 GCPs. I am displaying the table that shows the RMS error. In the GCP selection dialog, I can adjust the degree up to 3 (the number of GCPs determines how many degrees can be achieved) and the RMS error in the table updates. For example: Using first degree: ; ENVI Ground Control Points Table ; ENVI Image to Map GCP Table ; Map (x,y), Image (x,y), Predict (x,y), Error (x,y), RMS Error ; Total RMS Error: 1.873113 ; 477552.34 4433554.00 959.17 790.83 956.66 792.20 -2.51 1.37 2.86 477573.97 4440070.00 955.50 139.75 954.69 139.47 -0.81 -0.28 0.86 477524.00 4437949.00 950.25 350.75 951.04 351.96 0.79 1.21 1.45 Using degree 3: ; ENVI Ground Control Points Table ; ENVI Image to Map GCP Table ; Map (x,y), Image (x,y), Predict (x,y), Error (x,y), RMS Error ; Total RMS Error: 0.675043 ; 477552.34 4433554.00 959.17 790.83 958.10 791.36 -1.07 0.53 1.19 477573.97 4440070.00 955.50 139.75 955.52 139.68 0.02 -0.07 0.07 477524.00 4437949.00 950.25 350.75 950.38 351.06 0.13 0.31 0.34 Can you explain further what you are seeing? Or perhaps it would be best to work directly with Technical Support.
    You are not authorized to post a reply.