Scenario 1: Processing an individual SAR image exclusively in Object Analyst

You can use the same SAR image for Segmentation and Attribute Calculation. You can process SAR data in its raw vendor format or, alternatively, you can ingest it to a PCIDSK (.pix) file for processing.

Note: The SAR imagery you want to use for object classification must not have had any other processes run previously on it; that is, the metadata must be in its original state. Preprocessing, such as, for example, speckle filtering, can alter the metadata such that it cannot be processed by Object Analyst.

The input imagery can have detected and complex channels from single, dual, or fully polarimetric images.

If necessary, you can segment SAR data based on the image-covariance eigenvalues, L1 (1), L2 (2) and L3 (3). You can also, as an option, apply data averaging, tail trimming, or both on all image channels (HH, HV, VH, VV, RH, RL…) or any subset thereof to the eigenvalues.

After running Segmentation, you can then run Attribute Calculation. Object Analyst provides a set of predefined attribute types for your convenience.

Statistical and polarimetric attributes are calculated at the object level. All pixels within the boundaries of an object are averaged to a covariance ([C]) or coherency ([T]) matrix to calculate the polarimetric parameters. Attribute Calculation for texture is a two-stage process, as follows:

The selected texture parameters are generated temporarily based on a window size you specify (n x n).
The mean of each texture image is calculated for each object.
For more information about the predefined attributes, see Object attributes, calculation, and band aliases.

The following figure shows the workflow of using the same SAR image for Segmentation and Attribute Calculation.

Figure 1. Workflow using same SAR image

Workflow using same SAR image