RETRO commands


PROGRAM NAME

vmindis - Minimum Distance Classifier (K1)

DESCRIPTION

vmindis is a simple minimum distance classifier. The distance metric used to the Euclidean distance. This routine accepts two images as the input. The image that corresponds to the -i1 argument is the image that needs to be classified. This image can be have as many data bands as necessary. The number of data bands depicts the dimensionality of the vectors. Each pixel in the image is a vector with dimensionality of the number of data bands.

The other input image which corresponds to the -i2 argument is the prototype image. The last data band of this image is the class data band. This image must contain the same number of data bands as the other input image plus an extra data band that represents the class mapping. This image would most likely have been created by vkmeans or some other routine that will give cluster centers. This image contains vectors that correspond to the prototype of each class.

As stated above the center image's last data band is a class data band. The class data band simply maps each vector in the center image to the final class.

At this point most class images must be created manually. A class image can be created by using pseudo color in editimage on the resulting clustered image from the vkmeans routine. Vcustom can then be used to create the class image. Finally kinset can be used to add the class image data to the center image. Normally one would over cluster using vkmeans then reduce the space down. So vkmeans might produce 100 clusters but, really only 3 clusters are desired. Pseudo color in editimage and kasc2val can be used to reduce the space down and create a class image. The class data might map cluster center vectors 1-50 to class 1, vectors 51-60 to class 2 and vectors 61-100 to class 3. When the vmindis routine is run, the result would be a classified image of three classes.

The border option (-b) allows the user to specify a border width, in pixels, encompassing the image. The border region is skipped by vmindis when classification is performed. This useful if neighborhood operators have been used previously and have not updated edge pixels.

All input images must be of data storage type FLOAT.

REQUIRED ARGUMENTS

-i1
type: infile
desc: input image to be classified
-i2
type: infile
desc: input center/class image
-o
type: outfile
desc: output specifying which vector belongs to which cluster

OPTIONAL ARGUMENTS

-b
type: integer
desc: specifies the border width in pixels (default = 0)
default: 0
bounds: 0 < [-b] < 100

EXAMPLES

vmindis -i1 aerial_image.xv -i2 vkmeans_centers -b 5 -o test

aerial_image.xv is a 5 band image; thus, each vector has dimensionality of 5, vkmeans_centers image is a 6 band image, 5 bands are the cluster center values produced by vkmeans and 1 class band. The border that vmindis will ignore is 5 pixels wide. the output image will be a single band image called test.

SEE ALSO

RESTRICTIONS

All input images must be of data storage type FLOAT.

REFERENCES

COPYRIGHT

Copyright (C) 1993 - 1997, Khoral Research, Inc. ("KRI") All rights reserved.