How FME Processes Raster Features

Raster data in FME is represented by features with raster geometry.

Raster data differs in several key ways from vector data, and is handled uniquely in FME. A raster can be considered as a grid of values organized into rows and columns, with the relative size of its cells determining its resolution, or level of detail. Each row and column intersection in raster grid is called a cell or pixel. Vector point geometries can often be thought of as analogous to cells, while linear geometries like roads are represented as contiguous cells. Vector features tend to be more meaningful when taken together as a group, while a single raster feature can convey the same spatial information in a non- modular fashion.

Often a raster serves as a backdrop for overlaying specific vector information. For example, you can place vector lines and polygons that represent streets and buildings on a raster image that is an aerial photograph of a city. Conversely raster may be used as a backdrop for creating vector maps. For example, one might use satellite or aerial imagery to map particular features of an urban environment for city planning or geologic structures for locating natural resources.

Rasters can be represented as either image or numeric data.

Images are commonly derived from satellite data or photography, while numeric data often represents elevations, temperatures, and other quantitative information.

Raster data is stored in one or more bands whose properties may or may not be homogeneous.

Imagery data often contains several bands of data. This data may come from either Optical or SAR (Synthetic Aperture RADAR) type sensors and may contain any number of image bands of data relating to a wide range of spectral bands or polarizations . Although a band may have its own specific band properties, all bands on a raster must share a common set of raster properties such as the number of rows and columns, the cell size and the ground extents.

A band may optionally have one or more palettes, also called colormaps or Look-Up Tables (LUTs), associated with it.

A palette is essentially a lookup table of discrete keys to color or string values. These rasters are often referred to as classified because of the discrete data ranges. Classified rasters often serve the purpose of providing a visual representation or providing additional descriptive information for specific areas of a raster.