Segment Mean Shift function

Available with Image Analyst license.

Available with Spatial Analyst license.


Identify objects, features, or segments in your imagery by grouping adjacent pixels together that have similar spectral and spatial characteristics. You can control the amount of spatial and spectral smoothing to help derive features of interest.


The input raster needs to be 8 bits and either one or three bands.

For optimum segmentation results, it is recommended that you prepare your input raster layer to best discriminate your features of interest:

  • If your raster is more than three bands, specify the optimum band combination using the Extract Bands function.
  • Stretch your image to display your features of interest to maximum advantage using the Stretch function. If your raster data is other than 8 bits, use the Stretch function to specify 8 bit Unsigned as the Output Pixel Type in the General tab of the function.

The output layer from the preprocessing steps above is the input to the Segment Mean Shift function.

The classification training tools require that the input segmented raster dataset be a file. Persist your segmentation layer by clicking on Save As, and assign a file name for the segmented raster. The time to process the entire segmented raster dataset may be lengthy if the input layer is large.




The input raster to be segmented.

Spectral Detail

The relative importance of separating objects based on color characteristics.

Valid floating-point values range from 1.0 to 20.0. Smaller values result in broad classes and more smoothing. A higher value is appropriate when you want to discriminate between features having somewhat similar spectral characteristics. For example, using a higher spectral detail value in a forested scene allows you to better distinguish the different tree species.

Spatial Detail

The relative importance of separating objects based on spatial characteristics.

Valid integer values range from 1 to 20. Smaller values result in broad classes and more smoothing. A higher value is appropriate for discriminating between features that are spatially small and clustered together. For example, in an urban scene, you could classify general impervious surface features using a smaller spatial detail value, or you could classify buildings and roads as separate classes using a higher spatial detail value.

Minimum Segment Size in Pixels

The minimum segment size, measured in pixels. This value is related to your minimum mapping unit, and will filter out smaller blocks of pixels. All segments that are smaller than the specified value will merge the smaller segments with their best fitting neighbor segment.

Segment boundaries only

The segment boundaries draw as a black contour line around each segment. This is helpful so you can distinguish adjacent segments that have similar colors.

  • Unchecked—The segment boundaries are not displayed. This is the default.
  • Checked—The segment boundaries are displayed with black contour lines around each segment.

Maximum Segment Size in Pixels

The maximum size of a segment. Segments that are larger than the specified size will be divided. Use this parameter to prevent artifacts in the output layer resulting from large segments. The default is -1.

Related topics

In this topic
  1. Overview
  2. Notes
  3. Parameters