This algorithm converts thematic objects to sample objects. Between sample object polygons and image objects the conversion is based on an overlapping condition. For point vectors samples are created if all points within an image object are of the same class. This algorithm can be used for supervised classification and accuracy assessment.
Supported Domains
Image object level; Image object list
Algorithm Parameters
Thematic layer
Select the thematic layer to derive sample objects from.
Class name attribute
Thematic layer attribute that contains the class name.
Match class names
Match class hierarchy names to thematic layer class names. (In case the class names are exactly the same the matching is done automatically.)
Overlap requirement
Select how class for image object overlapping with thematic object is selected. If the overlap requirement is not met, no sample will be created. (Note - this parameter is ignored for point vectors).
Choose between:
-
largest overlap - An image object is converted to a sample object if there is an overlap and the thematic object with the largest overlap is taken.
-
threshold based - An image object is converted to a sample object if there is an overlap higher than a defined threshold in percentage. See parameter Overlap threshold.
In case different classes have an overlap for one image object, that class with the higher overlap is taken for the sample class.
Overlap threshold
Overlap threshold in percentage [0%...100%]. Only objects with single class overlap higher than specified threshold will be converted. The default value is higher than 50% overlap. (This parameter is only available for selection of Overlap requirement > threshold based and is ignored for point vectors.)
