Available with Image Analyst license.
With ArcGIS Image Analyst, you can classify and identify complex patterns in multispectral imagery and multidimensional raster data.
Image classification is the science and art of quantifying the identification of features or objects in imagery. It uses the spectral, shape, and spatial characteristics of objects in imagery to categorize them into classes. Supervised and unsupervised classification is supported, along with pixel and object-oriented approaches. Statistical, non-parametric, spectral matching, and machine learning classifiers are available. The training sample manager allows you to establish classification schemas, training sample collection, editing, and management of samples. You can assess the accuracy of training samples, and of resulting classification. You can use the classification wizard that guides you through the classification workflow, or use the suite of classification tools.
Regression tools are available for identifying, classifying, and quantifying patterns of ground features and phenomena. Specifically, the random forest regression model classifier tool accepts multiple multispectral images or multidimensional rasters as explanatory variables to describe a target dataset. If you use a multidimensional dataset, the dimensions are the exploratory variables, and can be assigned importance values that contribute to predicting the target variable. The output from the random forest regression model classifier tool can be used to generate a raster of predicted data values for modeling.
The classification and pattern recognition tools are closely associated with the tools and capabilities in the Change Detection, Deep Learning, and Multidimensional Analysis toolsets.
Classification Tools
The Classification Tools are available when an image is active in the Contents pane, and you click the Imagery tab. In the Image Classification group, click the Classification Tools drop-down list to display the tools for classifying imagery.
The Classification Wizard is also available in the Image Classification group.
Classification and Pattern Recognition geoprocessing tools and raster functions
A suite of geoprocessing tools is available in the Classification and Pattern Recognition toolset, available with ArcGIS Image Analyst.
Several of the tools are also available as raster functions. See the Classification section in List of raster functions to learn more.
Automation using ArcPy
You can perform automated image classification using the extended Raster object and the Image Analysis module in ArcPy.
The Raster object supports multispectral imagery, and many properties and functions provide management and analysis capabilities for classified data. Because the Raster object is temporary, you can perform on-the-fly analysis, avoid intermediate data creation and storage, and save your results when the analysis is complete.
The Image Analysis module (arcpy.ia) is an extension of the ArcPy Python site package, and it provides access to functions, classes, and operators that support classification and other advanced raster analysis. In addition to on-the-fly analysis, you can run classification geoprocessing functions with the Image Analysis module to generate new datasets in a Python script.