Classifying Object Based on Mean IR Values

Question

While Developing rule sets using e-Cognition for object classification for say; Landsat 7 images what are the threshold conditions to classify water, building, agriculture land & bare soil according to IR values? Why water have less IR values compared to others? Is that due to the factor that water have only reflectance below 0.8 as shown below in figure 1. In the same manner can vegetation and soil values of reflectance be assigned as between 0.6-1.4 (due to their high reflectance as shown in figure 2) & >1.6 respectively?

Answer 1

I don't quite understand what your question is? What you have written is basically correct. Different objects absorb and reflect sunlight differently at different wavelengths. In some cases, the different spectral properties of different objects are also called spectral fingerprints, maybe that will help you in your search. This phenomenon is a key principle in remote sensing and object classification, especially in pixel-based classification. In addition, there are so-called indices that use the spectral properties of different objects to make them easier to recognize. One example is the well-known NDVI, which is used to recognize vegetation. Please ask again if anything remains unclear.