trimmed_mean ( Image, Mask : ImageTMean : Number, Margin : )

Smooth an image with an arbitrary rank mask.

The operator trimmed_mean carries out a non-linear smoothing of the gray values of all input images (Image). The shift mask (Mask) is transmitted in the form of a region. The average of Number gray values located near the median is calculated. Several margin controls can be chosen for filtering (Margin):

    0...255   pixels outside of the image edges
              are assumed constant (with the indicated
              gray value).
    -1        Continuation of edge pixels.
    -2        Cyclic continuation of image edges.
    -3        Reflection of pixels at the image edges.
The indicated mask (= region of the mask image) is put over the image to be filtered in such a way that the center of the mask touches all pixels once. For each of these pixels all neighboring pixels covered by the mask are sorted in an ascending sequence according to their gray values. Thus, each of these sorted gray value sequences contains exactly as many gray values as the mask has pixels. If F is the surface of the mask the average of these sequences is calculated as follows: The first (F - Number)/2 gray values are ignored. Then the following Number gray values are summed up and divided by Number. Again the remaining (F - Number)/2 gray values are ignored.


Parameters

Image (input_object)
image(-array) -> object : byte
Image to be filtered.

Mask (input_object)
region -> object
Image whose region serves as filter mask.

ImageTMean (output_object)
image(-array) -> object : byte
Filtered output image.

Number (input_control)
integer -> integer
Number of averaged pixels. Typical value: Surface(Mask) / 2.
Default value: 5
Suggested values: 1, 3, 5, 7, 9, 11, 13, 15, 17, 19, 21, 23, 25, 27, 29, 31
Range of values: 1 <= Number <= 401
Minimum increment: 1
Recommended increment: 2

Margin (input_control)
integer -> integer
Margin control: 0...255 (constant), -1 (edge pixels continued), -2 (cyclic continuation), -3 (reflection).
Default value: -3
Range of values: -3 <= Margin <= 255


Example
read_image(:Image:'fabrik':) >
draw_region(:Region::) >
trimmed_mean(Image,Region:TrimmedMean:5,-3:) > 
disp_image(TrimmedMean:::).

Complexity

For each pixel: O(sqrt(F) * 5) with F = area of the mask.


Result

If the parameter values are correct the operator trimmed_mean returns the value TRUE. The behavior in case of empty input (no input images available) is set via the operator set_system(::'no_object_result',<Resultat>:). If necessary an exception is raised.


Possible Predecessors

read_image, draw_region, circle, rectangle1


Possible Successors

threshold__, dyn_threshold__, regiongrowing__


Alternatives

sigma__, median_weighted, median


See also

circle, rectangle1, min1__, max1__


References

R. Haralick, L. Shapiro; "Computer and Robot Vision"; Addison-Wesley, 1992, Seite 320



Copyright © 1996-1997 MVTec Software GmbH