Spatial Filters compute the intensity at a pixel
location by merging the values of neighboring pixels in a
particular way. The kind of filter determines the way the values
are merged. The pixels sampled for computing the new value are
taken from a neighborhood around or near the pixel. This type of
processing is used for enhancement, of image features, suppression
of noise, conversion to other value representations, and other
applications. The term "spatial filter" is used to differential
this type of processing in the spatial domain from filtering in the
frequency domain as by using FFTs.
|
Spatial Filters
|
|
Binomial
|
Applies a filter kernel having binomial weights.
This filter approaches a Gaussian filter shape for sizes beyond
5--7 pixels, and works very fast.
|
|
Gaussian
|
Applies a Gaussian smoothing kernel to the
image.
|
|
Rectangle
|
Applies a rectangular smoothing kernel to the
image.
|
|
Elliptical
|
Applies an elliptical smoothing kernel to the
image.
|
|
Block Average
|
Combines adjacent pixels by averaging their values
in blocks of a specified size. Each block of pixels becomes 1 pixel
in the output image.
|
|
Block Sum
|
Combines adjacent pixels by summing their values
in blocks of a specified size. Each block of pixels becomes 1 pixel
in the output image.
|
|
Custom Filter Kernel
|
Applies a filter of arbitrary size using weights
you specify.
|
|
Custom 3x3 Filter
|
Allows you to specify the weights of a 3x3 filter
kernel. This is a simplified version of the
Custom Filter Kernel command.
|
|
Cosmic Ray
|
Applies a contrast-based point source rejection
filter. This works well with bright, isolated bad pixels like those
resulting from cosmic ray hits and radioactive decay tracks. The
filter rejects deviant pixels that exceed an adjustable threshold
compared to the brightness of their neighborhood.
|
|
Clip High Values
|
Replaces all pixels above a threshold with a
specified value.
|
|
Clip Low Values
|
Replaces all pixels below a threshold with a
specified value.
|
|
Clip Pixel Values
|
Replaces all pixels above or below a threshold
with a specified value.
|
|
High Pass
|
Applies a "high pass" sharpening filter. The
Strength parameter controls the effect. A setting of 0 performs
almost no sharpening.
|
|
Despeckle
|
Removes isolated "speckles", or outlying pixel
values. This is usually applied to RGB images to remove color
channel noise.
|
|
Unsharp Mask
|
Applies an unsharp mask sharpening filter. The
Properties specify the blurring profile which is subtracted from
the image.
|
|
Maximum Entropy Deconvolution
|
Applies the maximum Entropy method to sharpen
image details while reducing noise.
|
|
Maximum
|
Replaces pixels with the maximum value of pixels
inside the local neighborhood.
|
|
Minimum
|
Replaces pixels with the minimum value of pixels
inside the local neighborhood
|
|
Median
|
Applies a Median filter to the image. This rejects
outlying pixel values, no matter how extreme, and tends to preserve
edges relatively well, but images can become "blocky" from using
too large a kernel. This filter is accessible both from the
Smoothing group and the Rank Filters group.
|
|
Rank
|
Replaces pixels with a value that is the specified
percentile rank of all pixels inside the filter kernel. The
minimum, maximum, and median filters are special cases of this
filter using ranks of 0, 100, and 50, respectively.
|
|
Compass Gradient
|
Applies a directional gradient filter. This type
of filter enhances details lying along the pointing ("compass")
direction of the filter.
|
|
Rotational Gradient
|
Applies a filter that mainly enhances structure
that winds outward from points like the spokes of a bicycle
wheel.
|
|
Laplacian
|
Applies a Laplacian operator to enhance changes in
slope or gradient. The actual gradient is discarded by the filter,
leaving a measure of the strength of edges and discontinuities of
image features. This filter is usually employed to trace edges of
features in the image.
|
Mira Pro x64 8.83 User's Guide, Copyright Ⓒ 2026 Mirametrics, Inc.
All Rights Reserved.
