All Packages  Class Hierarchy  This Package  Previous  Next  Index

Class image.PubFunc

java.lang.Object
   |
   +----image.PubFunc

public class PubFunc

extends Object

Author:

Di Zhong (Columbia University)

MAX_ITERATION

Maximum iterations in k-means clustering algorithm (to avoid dead loop)
Default value is 100

PubFunc()

add(float[], float[])

Add the second float vector to the first one

divide(float[], float)

divide float vector f1 by v

kmeans(float[][], int, float[], float[][], int)

k-means clustering algorithm (using weighted Euclidean distance)
The function returns k central vectors

mean(float[][])

Compute the mean of vectors The function returns mean vector

mean(Vector)

Compute the mean of vectors The function returns mean vector

median(float[])

Return the median value of a float array, i.e.

nearest(float[][], float[], float[])

Return index of the nearest vector in an array to the query one Using weighted Euclidean distance

quicksort(float[])

Quick sort algorithm for float array

quicksort(int[])

Quick sort algorithm for integer array

quicksort(Sortable[])

Quick sort algorithm for Sortable object array

randomSelect(float[][], int)

Randomly select k vectors from a data set The function returns k vectors

sqdist(float[], float[], float[])

Compute the squared Euclidean distance (weighted) between two float vectors

sqdist(float[], float[], float[], float)

Compute the squared Euclidean distance (weighted) between two float vectors

 public static int MAX_ITERATION

Maximum iterations in k-means clustering algorithm (to avoid dead loop)
Default value is 100

 public PubFunc()

 public static float[][] kmeans(float data[][],
                                int k,
                                float w[],
                                float seed[][],
                                int nsamples)

k-means clustering algorithm (using weighted Euclidean distance)
The function returns k central vectors

Parameters:

data - an array of feature vectors: number=data.length; dimension=data[0].length

k - number of desired classes

w - wighting vector (null for no-weighting)

seed - initial k seed (central) vectors (null for random selection)

nsamples - number of samples to be classified at each iteration (0: classify all vectors)

 public static final float[] mean(float data[][])

Compute the mean of vectors The function returns mean vector

Parameters:

data - an array of feature vectors: number=data.length; dimension=data[0].length

 public static final float[] mean(Vector data)

Compute the mean of vectors The function returns mean vector

Parameters:

data - an array of feature vectors: number=data.size(); dimension=data[0].length

 public static final float[][] randomSelect(float data[][],
                                            int k)

Randomly select k vectors from a data set The function returns k vectors

Parameters:

data - an array of feature vectors: number=data.length; dimension=data[0].length

k - number of desired vectors

 public static final int nearest(float data[][],
                                 float v[],
                                 float w[])

Return index of the nearest vector in an array to the query one Using weighted Euclidean distance

Parameters:

data - an array of feature vectors: number=data.length; dimension=data[0].length

v - query vector

w - weighting vector (null for no-weighting)

 public static final float sqdist(float f1[],
                                  float f2[],
                                  float w[])

Compute the squared Euclidean distance (weighted) between two float vectors

Parameters:

f1 - the first float vector

f2 - the second float vector

w - the wighting vector (null for no-weighting)

 public static final float sqdist(float f1[],
                                  float f2[],
                                  float w[],
                                  float th)

Compute the squared Euclidean distance (weighted) between two float vectors

Parameters:

f1 - the first float vector

f2 - the second float vector

w - the wighting vector (null for no-weighting)

th - if distance (squared) between f1 and f2 is larger or equal th, return 1e+20; otherwise return distance. (To speed up searhing, esp for high dimensio vector)

 public static final void add(float f1[],
                              float f2[])

Add the second float vector to the first one

Parameters:

f1 - the first float vector

f2 - the second float vector

 public static final void divide(float f1[],
                                 float v)

divide float vector f1 by v

Parameters:

f1 - the float vector

v - the divider

 public static final float median(float a[])

Return the median value of a float array, i.e. return (a.length/2)th element after sorting
Note: original array is modified

 public static final void quicksort(int a[])

Quick sort algorithm for integer array

 public static final void quicksort(float a[])

Quick sort algorithm for float array

 public static final void quicksort(Sortable a[])

Quick sort algorithm for Sortable object array

All Packages  Class Hierarchy  This Package  Previous  Next  Index