▸ Observer rating:
This technique is very useful in tracking urban sprawl, land degradation and desertification, forest degradation, river morphology, valleys and a range of other topics….
This technique is divided into two ways:
1) Observer rating
2) Unattended classification
In the controlled classification, the user or image analyzer “supervises” the pixel rating process. The user determines the different pixel values or spectral signatures that must be associated with each category. This is done by selecting representative sample locations for a pre-known category in nature.
These samples are called representative sites and the computer algorithm then uses spectral signatures from these representative sites to categorize the entire image; ideally, category classifications should not overlap for this and for the accuracy of the classification process the boundaries of sample areas should be somewhat far from the limits of the category in which the sample is taken.
In the controlled classification, the largest impairment occurs before the actual classification process. Once the rating is turned on, the result is an objective image with classified categories that correspond to the types of land cover.
Controlled classification can be more accurate than unattended classification, but it relies heavily on representative sample locations, individual image processing skill, and spectral class excellence.
If two or more categories are very similar to each other in terms of their spectral reflection (e.g., seasonal grasslands, perennial grasslands or shallow and deep waters), the likelihood of misclassification will increase.
Controlled classification requires considerable attention to the development of representative samples. If representative samples are not expressive or unrepresentative, the rating results will also be poor, so the generally controlled classification requires more time and money than the unattended classification.
▸ Observer rating:
1- First activate the Image classification window by right button on the clear empty area we have in the image indicated by the arrow and then activate it.
2-The Image classification window appears as follows:
This window may be not activated as in the form in this case we open the customize tab and then extensions to show us the next window where we activate spatial analyst.
3- We add polygon in places that represent a particular phenomenon where the program relies on these values to categorize the phenomena on this visual.
In this figure, the details of the shape appear.
polygon in the training sample manager window.
4- We add several forms covering the situations of the phenomenon that we have and then integrate them all using merge training samples under the name of the phenomenon as follows:
5-Thus, work is done to cover all the required areas:
6-Then we request the controlled classification from the classification list we choose Interactive Supervised Classification.
7-Our result shows the following: