Rochdi Posted November 4, 2020 Posted November 4, 2020 Hello Experts, After I merged S and L cubes into a single layer, I would like to represent these spectrums (Visible, NIR and SWIR) into a three dimensional cube. I used a "3D cube" tool in ENVI toolbox to do this, then I chose 3 bands from each mentioned spectrum to make a color composite.However, I didn't get the expected result as seen below. Is something wrong? I look forward to your reply. Rochdi.
Ray Arvidson Posted November 9, 2020 Posted November 9, 2020 (edited) Hi: I am here to help. Tell me what you mean by a 3D cube? Do you want a data cube where planes are images and each pixel has a spectrum associated with it? So when you display the S and L combined cube and use the spectral plot as you move around the image in the display the spectrum for that location shows in a plot window. Is that what you want? Build 3D cube does not do that and only produces a nice perspective view of the cube with an image on the top and color-coded spectral values down the side. https://www.l3harrisgeospatial.com/docs/ImageCubes.html Ray Arvidson Ah, I see the problem with 3D data cube. It has to be a rectangular cube, square or rectangular. Not a map projected version, see attached. Edited November 9, 2020 by Ray Arvidson added some text
Rochdi Posted November 10, 2020 Author Posted November 10, 2020 (edited) Hi Sir, Thanks a lot for your help, What I meant by building a 3D cube is exactly what you attached above, a 3D view of the cube with an image on the top and color-coded spectral values as well on the side. Depending on my system capability, it seems fine to build a 3D cube only for the VNIR range. Is it possible to build a 3D cube only for VNIR range relying on the instructions you mentioned in the link above? I followed the instructions in your mentioned link to come up with a 3D cube for VNIR range. I selected the input file and I performed a spectral subsetting of 107 bands. Then, I chose the bands I want to place on the face of the image cube (R: Band 51 | G: 34 | B: 24) and a Rainbow as color side. To apply a multiplier to the pixel dimensions of the side and top, I kept the border and spectral scale values as default. However, I come up with these unexpected output results as I attached below. Look forward to your reply, Rochdi. Edited November 10, 2020 by Rochdi
Ray Arvidson Posted November 10, 2020 Posted November 10, 2020 I suggest you look at the range of values in your cube. If there are extremely high excursions because of noise, or in L data, use of 65535 to indicate non-data, the system will scale its output relative to the extreme values and results will be black. Remove these values before proceeding. Ray
Rochdi Posted November 10, 2020 Author Posted November 10, 2020 Hello Sir Ray, I did a quick stats for the VNIR range (S cube) to check which bands contain the mentioned value, and I found the first five bands of this range with 65535 value. Then, I exclude these none-data using ROI and subset to come up with a sub image without 65535 value before building the 3D cube. Here's the outputs I came up with as I attached below (then, I edit the header file of the output by adding data ignore value = 0 in order to remove the black area that surround the cube). Would you please let me know whether the output is accurate or not? Rochdi.
Ray Arvidson Posted November 11, 2020 Posted November 11, 2020 Yes, your results are what they should be, so congratulations. Ray Arvidson
Rochdi Posted November 11, 2020 Author Posted November 11, 2020 My thanks and appreciation to your help Sir Ray Arvidson .
Recommended Posts
Create an account or sign in to comment
You need to be a member in order to leave a comment
Create an account
Sign up for a new account in our community. It's easy!
Register a new accountSign in
Already have an account? Sign in here.
Sign In Now