3136 Rate this article: No rating Potentially unexpected results when using spectral resampling with a user-defined filter function You may experience unexpected spectral resampling results when using a user-defined filter function. This Help Article discusses one possible solution to this problem. When trying to spectrally resample an image with a user-defined filter function, the resulting spectra may have a much lower spectral response than expected. This is caused by the way the filter function is defined and how ENVI interprets the 'end of the function response'. ENVI expects that a spectral response function has a definitive beginning and end where the response is 0.0 reflectance or radiance. If you try to perform spectral resampling with a filter function that does not reach 0.0 an the beginning and end of the response, ENVI will try to interpolate the function and it may interpolate to a negative response. This will cause the resulting spectra to have a much lower response than expected. To fix this problem, it is necessary to add additional wavelengths of 0.0 response on the beginning and end of the response function. This is most easily accomplished by exporting the spectral library to an ASCII file from the Plot window's File menu. The output ASCII file will look similar to the following: ENVI ASCII Plot File [Fri Jan 20 08:56:03 2006] Column 1: Wavelength Column 2: Spectral Response.txt 450.000000 0.000010 455.000000 0.000011 460.000000 0.000013 ... In a text editor, add two additional spectral wavelengths and response values of 0.0 at the beginning and ending of the ASCII response file. For example, at the beginning, it may look like the following: ENVI ASCII Plot File [Fri Jan 20 08:59:08 2006] Column 1: Wavelength Column 2: Spectral Response.txt 0.440000 0.000000 0.445000 0.000000 0.450000 0.000010 0.455000 0.000011 0.460000 0.000013 ... Save the ASCII file and import this using the Spectral Library Builder from ENVI's Spectral Menu (see the ENVI Application help for details on how to build a spectral library). Now the spectral filter function should have a definitive beginning and ending of 0.0 response. If used in spectral resampling, the results should be as expected. Review on 12/31/2013 MM Please login or register to post comments. How to combine the statistics of ROIs from different images to use in image classification RX (Reed-Xiaoli) Anomaly Detection in ENVI