This research project tested the quality of DIRSIG's texture characterization routines for a grass target and potential techniques to improve it. The objective of the DIRSIG is to produce output images that are analogous to real image outputs so the model can be applied to a wide range of problems. It is very important that both spectral and spatial properties of targets are modeled correctly to obtain realistic results. DIRSIG applies texture to each pixel spectrally. Current techniques examine only one bandpass region, using statistical means to compare a texture image in that bandpass with a spectral database in order to determine which curve will be used for a particular pixel. The research in this project examined the quality of this procedure, and several potential methods for improvement. DIRSIG images were generated to simulate two different scenes for which real images exist. ENVI’s principle components analysis utility was than used to quantify the information in each image for comparison. Modifications were than made to the process to expand the spectral database, and incorporate a new algorithm into DIRSIG that examines multiple bandpass regions when determining which spectra to use for a pixel. The output for each of these new scenarios was than tested for comparison with DIRSIG’s previous results, and the results obtained for the truth images. This research has quantitatively examined the ability of DIRSIG to replicate real world texture characteristics, specifically for a grass target. Research discovered a large gap between the amount of information that is contained in images of real world data and those simulated by DIRSIG. Expanding the spectral database did produce a slight increase in results, increasing the qualitative appearance of the image as well. The amount of data contained in the DIRSIG image remained significantly less than that contained in the real image. Increasing the number of bandpasses employed by DIRSIG in determining which spectra to map onto a particular pixel, results indicated two different concepts. Adding one band increased the ability of DIRSIG to map correctly choose the spectra. Additional bandpasses may increase the quality of an image, however, choosing too many bandpass regions for an image can cause the resulting quality of the image to decrease. The process is very complex, the output quality is dependent on the particular imaging system that is being modeled, the bandpasses that are chosen, and the amount of bandpasses that are used as references.
Burtner, Christy, "Texture characterization in DIRSIG" (2001). Thesis. Rochester Institute of Technology. Accessed from
RIT – Main Campus