endrov.flowMultiscale
Class Multiscale

java.lang.Object
  endrov.flowMultiscale.Multiscale

public class Multiscale
extends java.lang.Object

Tools for working with MRA/MSA

Constructor Summary

Multiscale()

Method Summary

static double	convolveGaussPoint2D(EvPixels p, double sigmaX, double sigmaY, double midx, double midy) Convolve with a gaussian at a single point
static double	findFeatureScale(EvPixels p, double sigmaGuess, int x, int y) Find scale of feature by maximizing RickerWavelet(sigma) at a given point.
static void	findFeatureScale(EvPixels p, double sigmaGuess, int x, int y, java.util.Map<java.lang.Double,java.lang.Double> dog)
static double	findFeatureScale2(EvPixels p, int x, int y, double minSigma, double maxSigma, int sigmaDiv, int numIt) Find scale of feature by maximizing RickerWavelet(sigma) at a given point.

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Constructor Detail

Multiscale

public Multiscale()

Method Detail

findFeatureScale

public static double findFeatureScale(EvPixels p,
                                      double sigmaGuess,
                                      int x,
                                      int y)

Find scale of feature by maximizing RickerWavelet(sigma) at a given point. Uses DoG approximation.

Uses several 2^-series around the guessed sigma, supposedly very fast

findFeatureScale

public static void findFeatureScale(EvPixels p,
                                    double sigmaGuess,
                                    int x,
                                    int y,
                                    java.util.Map<java.lang.Double,java.lang.Double> dog)

findFeatureScale2

public static double findFeatureScale2(EvPixels p,
                                       int x,
                                       int y,
                                       double minSigma,
                                       double maxSigma,
                                       int sigmaDiv,
                                       int numIt)

Find scale of feature by maximizing RickerWavelet(sigma) at a given point. Uses DoG approximation.

Tries several sigma (sigmaDiv) within a range, then recursively looks within subintervals, numIt times.

convolveGaussPoint2D

public static double convolveGaussPoint2D(EvPixels p,
                                          double sigmaX,
                                          double sigmaY,
                                          double midx,
                                          double midy)

Convolve with a gaussian at a single point