Available with Spatial Analyst license.
Understanding and being familiar with the following terms will be helpful when performing image classification using the capabilities of the ArcGIS Spatial Analyst extension:
A raster dataset produced by scanning a surface with an optical or electronic device. Common examples include scanned documents, remotely sensed data (for example, satellite images), and aerial photographs. An image is stored as a raster dataset of binary or integer values that represent the intensity of reflected light, heat, or other range of values on the electromagnetic spectrum.
A cell is the smallest unit of information in raster data. Each cell represents the numeric value of some measure at the corresponding unit area location on the earth.
Cells are typically square in shape. The area each cell represents is dependent on the resolution of the raster. High-resolution (large-scale) raster cells would represent small areas, measured in units as small as square meters. The cells in a lower-resolution (small-scale) raster represent the uniform value of a larger area, such as hectares or square kilometers.
The smallest unit of information in an image or raster map, usually square or rectangular. The term pixel is often used synonymously with cell.
The process of sorting or arranging pixels in an image into classes or clusters. Depending on the interaction between the analyst and the computer, there are two types of image classification—supervised classification and unsupervised classification.
An image classification approach that is based on the training samples collected by the analyst. The training samples "teach" the software how to classify the rest of the pixels in the image.
An image classification approach that sorts the pixels in the image into clusters without the analyst's intervention. The process is based solely on the distribution of pixel values in a multidimensional attribute space.
A group of pixels in an image that represent the same object on the surface of the earth.
A group of pixels that is distinguishable in a multidimensional attribute space. A cluster is similar to a class except that the ground object that it represents is unknown when the clustering analysis is performed.
Sample areas in an image that represent different classes in a supervised classification. Training samples provide examples for the classes in an image, so that the classification tools know how to classify the rest of the pixels.
A signature file records the spectrum signatures of different classes across a series of bands. For each class, the signature contains means and covariances calculated from its training sample.
It would also be helpful to be familiar with general raster data vocabulary.