Core Optical, Inc.
AlphaTrimmedMeanFilter ClassIndustrial Strength, Scientific Grade
Encapsulates the functionality associated with an Alpha-Trimmed Mean Filtering operation.
Inheritance Hierarchy

OnlineSystem Object
  PrecisionImage.SpatialFiltering SpatialFilteringOperator
    PrecisionImage.SpatialFiltering AlphaTrimmedMeanFilter

Namespace: PrecisionImage.SpatialFiltering
Assembly: PrecisionImage (in PrecisionImage.dll) Version: (

public class AlphaTrimmedMeanFilter : SpatialFilteringOperator

The Alpha-Trimmed Mean filter is a spatial filter with characteristics of both a mean and a median filter. The values within a spatial neighborhood (defined by the kernel radius) are rank ordered by ascending value and the average is computed. High and low ordered values are "trimmed" from the working set via a filter parameter prior to the calculation of the average. If no values are trimmmed, the filter reduces to a mean filter. If the trim value is equal to half the kernel size (KernelRows * KernelColumns) / 2, the filter becomes a median filter. Trim values between these two extremes causes the filter to exhibit characteristics between those of a median and a mean filter. because of this, the Alpha-Trimmed Mean filter is usually most effective on datasets that exhibit both impulse and centered additive noise characteristics.

Use the TrimValue property to adjust the number of values to trim from the ends of the ranking. This value must be a positive value no greater than (KernelRows * KernelColumns) * 0.5 - 0.5, where the kernel Rows/Columns are determined via the KernelRadius property. The default value for TrimValue and KernelRadius is 1.

See Also