X
PrevPrev Go to previous topic
NextNext Go to next topic
Last Post 16 Sep 2011 03:08 AM by  anon
Making attribute map in ENVI plus conversion of vector map to raster
 1 Replies
Sort:
You are not authorized to post a reply.
Author Messages

anon



New Member


Posts:
New Member


--
16 Sep 2011 03:08 AM
    Dear all I tried to produce an attribute map in Envi 4.5 from the shape file imported from ArcGIS but to no avail. The other issue is pertaining to exporting vector to raster format in ENVI environment (Other than ROI). Is it possible? and the final query is what file format (excluding satellite images) does ENVI support to carry out supervise classification. The reason being that when I tried ESRI grid format (1 band), the message pops up....more than 1 band required. My intention is to produce a soil map based on several variable or soil forming factor maps produced in vector formats in ArcGIS (converted to raster) using artificial neural network (supervise classification) within ENVI. I did tried with the tiff format by stacking all the variable maps together, trained ANN and executed the classification. The only problem now is that I cannot label the attribute as can be done in the ArcGIS vector format (eg. paddy field..orchard..forset etc). Although I can see different polygons but cannot label the attribute. The attributes are just coded with RGB value (0-255). My worry is that, it may not represent soil forming factors during analysis. For instance if I re-export the mentioned variable maps again, the RGB value will changed based on the colour I use for each attributes. Any help to resolve the mentioned problem shall be highly appreciated. Regards Dorji

    MariM



    Veteran Member


    Posts:2396
    Veteran Member


    --
    21 Sep 2011 09:40 AM
    ENVI will not create an 'attribute map'. This would best be done in ArcGIS. You could create a classification type map where each 'attribute' is a class. As you have found, you would need to export the vectors to ROIs and then you can create a classification image from the ROIs. Many classification algorithms work on spectral information from more than one band, so you would need a multiband file to perform the classification. Some algorithms can work on single band files - Minimum Distance, Parallelepiped, SVM. The Unsupervised methods can also be used. Raster images do not contain attributes other than the information of the pixel. However, once you create the classification image, you could export each class to a vector where you could add attributes.
    You are not authorized to post a reply.