In Object Analyst, a basic workflow of object-based classification is shown in the following figure. You start by segmenting an image into objects. The idea is to partition the image into statistically homogenous regions or objects (segments) that are more uniform within themselves and differ from their adjacent neighbors.
Next, you select the attributes you want to calculate, which can be statistical, geometrical, Vegetation Indices, and more. After calculating the attributes, you can run either a supervised or unsupervised classification, and refine the classification by applying one or more rule-based classifications you define.
You can use a variety of post-classification editing tools to reform shapes, modify classes, and more. With a supervised classification, you can also, if necessary, run a batch classification and an accuracy assessment.
