Full text: Mesures physiques et signatures en télédétection

586 
where n is the number of cover types in the restricted data set, R is the true inference for wavelength X 
and measurement i, and /? is the inferred inference for wavelength X and measurement i. Measurement i 
represents a specific surface with a specific sun angle. Using this statistic VEG provides a rigorous estimate 
of the accuracy of its inferences. 
Finally, VEG outputs the basic information about the goal, input data, techniques applied, and the 
inferred results and associated error bounds. 
23. Overview of VEG 
VEG is an object system with three basic layers: the Common Lisp or programming language layer, the 
KEE (Knowledge Engineering Environment, IntelliCorp) layer, and the knowledge base layer. All methods 
were designed and compiled in the Lisp layer. All objects and methods for both the VEG system and the 
VEG interface were encapsulated in the KEE layer. The database management system for managing histori 
cal data was constructed in the KEE layer. The knowledge base layer contains objects that are frames with 
complex slot structures that can hold values, hold function definitions as procedural attachments, hold graph 
ics or window objects, and can enforce constraints on what values they can contain. KEE is a mature 
development platform that supports a number of well engineered components including inference engines, 
windows, graphics tools, objects and inheritance, procedural attachments and other support needed for proto 
typing expert systems. KEE facilitates the implementation of state of the art AI components which, in turn, 
provide a flexible development environment. 
The VEG system uses several knowledge bases. Databases and a toolbox that are common to all parts of 
the VEG system, and the methods, techniques and rules for estimating vegetation parameters and estimating 
atmospheric effects are stored in the VEG knowledge base. The toolbox is available to the user whenever 
VEG is loaded. It can be used to browse any loaded knowledge base, print the screen, help the user, and 
plot target and historical data. 
VEG is embedded in an extensive, window oriented interface system that provides a variety of screens to 
enhance dialogue between the scientist and the system. The interface is a key feature of this system. It was 
designed to focus the scientist on the appropriate level of organization to carry out scientific work without 
attention to "housekeeping" functions. This interface allows the scientist to run VEG and select options at all 
stages of a session by clicking the mouse over the appropriate menu option. The interface allows a scientist 
with no knowledge of KEE, Common Lisp or the detailed structure of VEG to use the system with ease. 
The only time that the scientist needs to use the keyboard during a session is if he/she chooses to enter new 
data manually. All other operations are controlled by the mouse. The interface allows the scientist to focus 
on the data and the functions performed by VEG, and it abstracts away most of the underlying detailed com 
plexity of the VEG system. 
VEG has two modes of treating target data: the "Automatic Mode" and "Research Mode". The 
Automatic Mode is used when the scientist would like to process a large number of unknown targets (e.g. 
from a large region/scene of the Earth). The Research Mode is used when the scientist is exploring a few 
unknown targets or developing and/or testing inference techniques on the internal historical data base. The 
Research Mode gives the user the full power and flexibility of the system. An interface to an input file of 
unknown cover type data is available when VEG is operating in the "Automatic Mode." The interface 
enables the user to name the input file and specify the format for the file. The format can be a standard, 
previously-defined or newly-defined faro at. Using this format, the input file is read and the cover type data 
are stored for processing in the system. Formats that have been newly defined using the interface can be 
stored for subsequent use. The output from both modes can be written to a file using a standard, 
previously-defined or newly-defined format. 
The toolbox is an important part of VEG. The user can activate the toolbox at any time during a ses 
sion. The toolbox allows the user to read a description of the VEG system; browse the objects and slots 
within the VEG system; obtain help about any screen; plot the zeniths, azimuths and reflectance values of 
reflectance data in two different plots; explore the historical database; and print out a screen dump of the 
current screen. The toolbox provides a means of managing the levels of abstraction the scientist sees and 
allows the scientist to deepen his understanding of the system functionality if he wishes to do this. 
A simple interface that allows a scientist to add new techniques, e.g., estimating leaf area index without 
having knowledge of KEE and having a detailed knowledge of the structure of VEG has been implemented.
	        
Waiting...

Note to user

Dear user,

In response to current developments in the web technology used by the Goobi viewer, the software no longer supports your browser.

Please use one of the following browsers to display this page correctly.

Thank you.