Alavipanah, Seyedkazem 
  
1- Any unwanted disturbance in image data that is due to limitation in sensing, signal 
digitization, or data missing were studied and noisy pixels were masked for further 
analysis, 
2- The initial statistics and histograms were extracted for evaluation of information 
content of TM bands finding pixels with particular Digital Numbers (DN) assessing 
the radiometric quality of the images and also peaks and valleys correspond with types 
of crusts. In this study, the Principal Component Analysis ( PCA ) is performed based 
on correlation matrix, 
3- The TM FCC,S and Photomorphic Unit Analysis ( PMUA ) and ancillary data were 
used to improve the image interpretation in a visual way. The collection of field data 
and use of accurate field observations were assisted in validity of the image analysis 
and selection of training areas, 
4- The validity of the training data is evaluated both from visual examination and from 
quantitative characterization. Therefore, in this study, the spectral signatures for salt 
crust types were evaluated by two dimensional feature space ( FS ) and statistics. The 
spectral signatures of the training samples was evaluated by using 7 TM bands. 
  
00 TM imagery, 7 Bands 
p Ne 
Derivation of Enhancemet of 
Statistics visual interpretation 
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
v 
  
Field work and training area 
  
  
  
  
  
  
  
Ancillary data y FS analysis 
TN 
  
  
  
  
Conclusion 
  
  
  
  
  
Figure 1, Flowchart indicates the mothodoloy of the research 
Result and discussions 
The result of tabulating frequencies of brightness values ( DN ) within the TM band 
1 shows that values of 255 correspond with salt crusts, eroded soil and desert crusts. 
The Desert crust areas are mainly non saline soil with bare soil surface conditions with 
a few millimeters thick and usually a bright surface. These areas located almost near 
  
40 International Archives of Photogrammetry and Remote Sensing. Vol. XXXIII, Part B7. Amsterdam 2000.