Application: gvSIG desktop » gvSIG geoprocess

<?xml version='1.0' encoding='ISO-8859-1' standalone='yes' ?>

        <help>

                <element name="DESCRIPTION" text="This algorithm generates one thematic layer (classified) from a group of images, clasifying its values in groups (classes). All of these images should contain information of the same zone using different  parameters. The most common practice for this is to use different bands of a  remote multispectral sensor although this is not the only possibility.&#10;&#10;The module should be provided with information previously known about a part of  the input data so that it can automatically apply it over the rest of the area being analysed. In short, helping the module  by showing it some areas (called training areas) that are known beforehand and share common characteristics with the area being analysed (homogenous class). These characteristics are known as spectral signature of each class and this signature is obtained from a series of statistical analysis of the values found in each training area and in each input layer. Once this signature is known, it is used to assign values to other cells that have similar signatures. The following figure shows a diagram of the process of supervised classification:" description="Description" type="0">

                        <image description="" file="clasSuper.jpg">

                        </image>

                </element>

                <element name="ADDITIONAL_INFO" text="" description="Additional information" type="0">

                </element>

                <element name="EXTENSION_AUTHOR" text="Victor Olaya" description="Extension created by" type="0">

                </element>

                <element name="HELP_AUTHOR" text="Victor Olaya" description="Help file created by" type="0">

                </element>

                <element name="USER_NOTES" text="" description="User_notes" type="0">

                </element>

                <element name="INPUT" text="one or various layers with their corresponding bands, through the multiple selection window." description="Bands" type="3">

                </element>

                <element name="POLYGONS" text="layer with training areas." description="Polygons" type="3">

                </element>

                <element name="FIELD" text="the field that contains the name of the class represented by each polygon." description="Field with class value" type="3">

                </element>

                <element name="METHOD" text="The assignment of values in the cells outside of the training areas also depend on the criteria used. This algorithms incorporates three of them:&#10;* Paralelepiped&#10;* Minimum distance&#10;* Maximum likelihood&#10;&#10;The user should check which of the algorithms produces the best classification of the image (this  depends on the  training areas as well as the characteristics of the images itself), checking some cells to see if their assigned values are correct. " description="Method" type="3">

                </element>

                <element name="OUTPUT_DESCRIPTION" text="" description="Description" type="2">

                </element>

                <element name="CLASSES" text="New attributes table called classes with statistical information for each one of the bands used." description="Classes" type="2">

                </element>

                <element name="CLASSIFICATION" text="the output raster." description="Classification" type="2">

                        <image description="input image" file="clasSuper_landsat_b4.jpg">

                        </image>

                        <image description="Paralelep&#237;ped" file="clasSuper_paralelepipedo.jpg">

                        </image>

                        <image description="Minimum distance" file="clasSuper_distancia_minima.jpg">

                        </image>

                        <image description="Maximum likelihood" file="clasSuper_maxima_similitud.jpg">

                        </image>

                </element>

        </help>

<?xml version='1.0' encoding='ISO-8859-1' standalone='yes' ?>