Batch Processing

As for the version 1.39 it is possible to run InterIMAGE in batch mode. With this new operation mode it is possible to create an interpretation project and then run it several times without having to interact with the user interface. This was mainly done to allow the execution of the so called pipelines. So, for example, now it is possible to run a classification project and a shape regularization of the building polygons found in the classification by writing only two command lines.

This is also interesting to test a classification project on different images or using different ROIs without having to change them using the interface. In fact, it is possible to write a batch file that executes different InterIMAGE instances to classify hundreds of images with little user interaction.

Finally, using some options it is possible to run slightly different configurations of the same operator or decision rule (this could be used to tune automatically the parameters values).

To execute InterIMAGE in batch mode it's necessary to create a project through the user interface. After that one should call it from the command line with at least these two following parameters:

interimage outputDecisionRule=“<decision_rule_file>” outputShapeFile=“<shape_file>”

It’s possible to use all these parameters:

outputDecisionRule=“<decision_rule_file>”

This decision rule allows the user to filter the results and compute features that will be stored in the exported shapefile. It is possible to create the decision rule file (*.dt) during the creation of the project.

outputShapeFile=“<shape_file>”

This parameter defines where the result shapefile (*.shp) will be saved.

semanticNetwork=“<semantic_network_file>”

The semantic network file (.net) is the one that stores the classes that are expected to be found in a scene and also the operators that will be executed during the interpretation. With this option the original semantic network will be replaced. It is possible, for example, to interpret a scene from different point of views (land cover / land use).

resourceImage=“<resource_name#image_file>”

With this option it is possible to change the images that are used in an interpretation. Resource name is the key given to an image resource in InterIMAGE.

resourceShape=“<resource_name#shape_file>”

With this option it is possible to change the shapefiles that are used in an interpretation. Resource name is the key given to a shape resource within InterIMAGE.

parameterOperator=“<@parameter_value_mask@#value>”

With this option it is possible to change the value of an operator parameter. Note that parameter_value_mask must be enclosed by @s. The parameter mask values must be unique for each operator parameter. With this option it is possible to run slightly different configurations of the same operator (this could be used to tune automatically the operator parameters values).

Example: interimage … “parameterOperator=@scale@#20”

In this case the operator parameter value should be set to @scale@ in the interface. When the project is executed in batch, the system will replace the parameter value accordingly.

parameterDecisionRule=“<parameter_value_mask#value>”

With this option it is possible to change the value of a decision rule parameter. Note that in this case the parameter value masks must be unique and ordered numbers in this format 00001, 00002 etc. With this option it is possible to run slightly different configurations of the same decision rule (this could be used to tune automatically the decision rule parameters values).

Example: interimage … “parameterDecisionRule=00001#75”

In this case the decision rule parameter value should be set to 00001 in the interface. When the project is executed in batch, the system will replace the parameter value accordingly.

It's possible to run an interpretation on a different image.

interimage “projectFile=project_name\project_name.gap” “outputDecisionRule=decision_rule.dt” “outputShapeFile=results.shp” “resourceImage=image_key#new_image”

• project_name is the name of your project;
• results is the shapefile that will be created with the resulting polygons;
• The decision_rule file should be created at the creation of the project;
• image_key is the key given to the image when the project was created;
• new_image is the new image file path.

It's possible to run an interpretation using a different shapefile. If the shapefile is used in a project to define a ROI, this parameter could be used to interpret a different ROI.

interimage “projectFile=project_name\project_name.gap” “outputDecisionRule=decision_rule.dt” “outputShapeFile=results.shp” “resourceShape=shape_key#new_shape”

• project_name is the name of your project;
• results is the shapefile that will be created with the resulting polygons;
• The decision_rule file should be created at the creation of the project;
• shape_key is the key given to the shapefile when the project was created;
• new_shape is the new shapefile path.

It's possible to run an interpretation using a different semantic network. The semantic network holds the information about the classes and their corresponding operators. So by changing the semantic network it's possible to do an interpretation according to different classes and run different operators.

interimage “projectFile=project_name\project_name.gap” “outputDecisionRule=decision_rule.dt” “outputShapeFile=results.shp” “semanticNetwork=new_semantic_network”

• project_name is the name of your project;
• results is the shapefile that will be created with the resulting polygons;
• The decision_rule file should be created at the creation of the project;
• new_semantic_network is the new semantic network path.

It's possible to run an interpretation changing the values of specific parameters in the interpretation project. With these options it is possible to run slightly different configurations of the same operator or decision rule (this could be used to tune automatically the parameters values).

interimage “projectFile=project_name\project_name.gap” “outputDecisionRule=decision_rule.dt” “outputShapeFile=results.shp” “parameterOperator=@scale@#20” “parameterDecisionRule=00001#75”

• project_name is the name of your project;
• results is the shapefile that will be created with the resulting polygons;
• The decision_rule file should be created at the creation of the project;
• @scale@ is an operator parameter value mask set at the creation of the project;
• 00001 is a decision rule parameter value mask set at the creation of the project.