gov.nih.mipav.model.algorithms
Class AlgorithmEdgeNMSuppression

java.lang.Object
  java.lang.Thread
      gov.nih.mipav.model.algorithms.AlgorithmBase
          gov.nih.mipav.model.algorithms.AlgorithmEdgeNMSuppression

All Implemented Interfaces:

AlgorithmInterface, ActionListener, WindowListener, Runnable, EventListener

public class AlgorithmEdgeNMSuppression
extends AlgorithmBase
implements AlgorithmInterface

Calculates the non-maximum suppression of an image at a scale defined by the user. This algorithm produces an edge map of the zero crossings of the non-maximum suppression for 2D images and 2.5D images. Edges are defined as the union of points for which the gradient magnitude assumes a maximum in the gradient direction. Introuduce a local orthonormal coordinate system (u,v) at any point P0, where the v-axis is parallel to the gradient direction at P0, and the u-axis is perpindicular. Iu = sin(a)*Ix - cos(a)*Iy Iv = cos(a)*Ix + sin(a)*Iy cos(a) = Ix/(sqrt(Ix*Ix + Iy*Iy)) evaluated at P0 sin(a) = Iy/(sqrt(Ix*Ix + Iy*Iy)) evaluated at P0 Iu = (Ix*Iy - Ix*Iy)/(sqrt(Ix*Ix + Iy*Iy)) = 0 Iv = (Ix*Ix + Iy*Iy)/(sqrt(Ix*Ix + Iy*Iy)) = sqrt(Ix*Ix + Iy*Iy) Iv*Iv = Ix*Ix + Iy*Iy Ivv = (cos(a)*Ix + sin(a)*Iy)(cos(a)*Ix + sin(a)*Iy) = cos(a)*cos(a)*Ixx + 2*cos(a)*sin(a)*Ixy + sin(a)*sin(a)*Iyy = (Ix*Ix*Ixx + 2*Ix*Iy*Ixy + Iy*Iy*Iyy)/(Ix*Ix + Iy*Iy) Iv*Iv*Ivv = Ix*Ix*Ixx + 2*Ix*Iy*Ixy + Iy*Iy*Iyy Ivvv = (cos(a)*cos(a)*Ixx + 2*cos(a)*sin(a)*Ixy + sin(a)*sin(a)*Iyy)(cos(a)*Ix + sin(a)*Iy) = cos(a)*cos(a)*cos(a)*Ixxx + 3*cos(a)*cos(a)*sin(a)*Ixxy + 3*cos(a)*sin(a)*sin(a)*Ixyy + sin(a)*sin(a)*sin(a)*Iyyy = (Ix*Ix*Ix*Ixxx + 3*Ix*Ix*Iy*Ixxy + 3*Ix*Iy*Iy*Ixyy + Iy*Iy*Iy*Iyyy)/((Ix*Ix + Iy*Iy)**3/2) Iv*Iv*Iv*Ivvv = Ix*Ix*Ix*Ixxx + 3*Ix*Ix*Iy*Ixxy + 3*Ix*Iy*Iy*Ixyy + Iy*Iy*Iy*Iyyy Assuming that the second and third-order directional derivatives of I in the v-direction are not simultaneously zero, a necessary and sufficient condition for P0 to be a gradient maximum in the gradient direction may be stated as: Ivv = 0, Ivvv < 0. Since only the sign information is important, this condition can be restated as: Iv*Iv*Ivv = 0, Iv*Iv*Iv*Ivvv < 0. Reference: Geometry-Driven Diffusion in Computer Vision, Bart M. ter Haar Romeny(Ed.), Chapter2:Linear Scale-Space II: Early Visual Operations by Tony Lindeberg and Bart M. ter Haar Romeny, Kluwer Academic Publishers, 1994, page 45.

For 3D a similar derivation: Iv = cos(a)*sin(b)*Ix + sin(a)*sin(b)*Iy + cos(b)*Iz where spherical coordinates are being used cos(a) = Ix/(sqrt(Ix*Ix + Iy*Iy)) evaluated at P0 sin(a) = Iy/(sqrt(Ix*Ix + Iy*Iy)) evaluated at P0 cos(b) = Iz/(sqrt(Ix*Ix + Iy*Iy + Iz*Iz)) evaluated at P0 sin(b) = sqrt(Ix*Ix + Iy*Iy)/(sqrt(Ix*Ix + Iy*Iy + Iz*Iz)) evaluated at P0 Iv = (Ix*Ix + Iy*Iy + Iz*Iz)/(sqrt(Ix*Ix + Iy*Iy + Iz*Iz)) = sqrt(Ix*Ix + Iy*Iy + Iz*Iz) Iv*Iv = Ix*Ix + Iy*Iy + Iz*Iz Ivv = (cos(a)*sin(b)*Ix + sin(a)*sin(b)*Iy + cos(b)*Iz)(cos(a)*sin(b)*Ix + sin(a)*sin(b)*Iy + cos(b)*Iz) = cos(a)*cos(a)*sin(b)*sin(b)*Ixx + 2*cos(a)*sin(a)*sin(b)*sin(b)*Ixy + 2*cos(a)*cos(b)*sin(b)*Ixz + sin(a)*sin(a)*sin(b)*sin(b)*Iyy + 2*sin(a)*cos(b)*sin(b)*Iyz + cos(b)*cos(b)*Izz = (Ix*Ix*Ixx + 2*Ix*Iy*Ixy + 2*Ix*Iz*Ixz + Iy*Iy*Iyy + 2*Iy*Iz*Iyz + Iz*Iz*Izz)/ (Ix*Ix + Iy*Iy + Iz*Iz) Iv*Iv*Ivv = Ix*Ix*Ixx + 2*Ix*Iy*Ixy + 2*Ix*Iz*Ixz + Iy*Iy*Iyy + 2*Iy*Iz*Iyz + Iz*Iz*Izz Ivvv = (cos(a)*cos(a)*sin(b)*sin(b)*Ixx + 2*cos(a)*sin(a)*sin(b)*sin(b)*Ixy + 2*cos(a)*cos(b)*sin(b)*Ixz + sin(a)*sin(a)*sin(b)*sin(b)*Iyy + 2*sin(a)*cos(b)*sin(b)*Iyz + cos(b)*cos(b)*Izz)(cos(a)*sin(b)*Ix + sin(a)*sin(b)*Iy + cos(b)*Iz) = cos(a)*cos(a)*cos(a)*sin(b)*sin(b)*sin(b)*Ixxx + 3*cos(a)*cos(a)*sin(a)*sin(b)*sin(b)*sin(b)*Ixxy + 3*cos(a)*cos(a)*cos(b)*sin(b)*sin(b)*Ixxz + 3*cos(a)*sin(a)*sin(a)*sin(b)*sin(b)*sin(b)*Ixyy + 6*cos(a)*sin(a)*cos(b)*sin(b)*sin(b)*Ixyz + 3*cos(a)*cos(b)*cos(b)*sin(b)*Ixzz + sin(a)*sin(a)*sin(a)*sin(b)*sin(b)*sin(b)*Iyyy + 3*sin(a)*sin(a)*cos(b)*sin(b)*sin(b)*Iyyz + 3*sin(a)*cos(b)*cos(b)*sin(b)*Iyzz + cos(b)*cos(b)*cos(b)*Izzz = (Ix*Ix*Ix*Ixxx + 3*Ix*Ix*Iy*Ixxy + 3*Ix*Ix*Iz*Ixxz + 3*Ix*Iy*Iy*Ixyy + 6*Ix*Iy*Iz*Ixyz + 3*Ix*Iz*Iz*Ixzz + Iy*Iy*Iy*Iyyy + 3*Iy*Iy*Iz*Iyyz + 3*Iy*Iz*Iz*Iyzz + Iz*Iz*Iz*Izzz)/ ((Ix*Ix + Iy*Iy + Iz*Iz)**1.5) Iv*Iv*Iv*Ivvv = Ix*Ix*Ix*Ixxx + 3*Ix*Ix*Iy*Ixxy + 3*Ix*Ix*Iz*Ixxz + 3*Ix*Iy*Iy*Ixyy + 6*Ix*Iy*Iz*Ixyz + 3*Ix*Iz*Iz*Ixzz + Iy*Iy*Iy*Iyyy + 3*Iy*Iy*Iz*Iyyz + 3*Iy*Iz*Iz*Iyzz + Iz*Iz*Iz*Izzz

Nested Class Summary

Nested classes/interfaces inherited from class java.lang.Thread

Thread.State, Thread.UncaughtExceptionHandler

Field Summary

static int	MARCHING_SQUARES Perform zero crossing detection using the marching squares method.
static int	OLD_DETECTION Perform zero crossing detection using Matt's old method.

Fields inherited from class gov.nih.mipav.model.algorithms.AlgorithmBase

destFlag, destImage, image25D, mask, maxProgressValue, minProgressValue, multiThreadingEnabled, nthreads, progress, progressModulus, progressStep, runningInSeparateThread, srcImage, threadStopped

Fields inherited from class java.lang.Thread

MAX_PRIORITY, MIN_PRIORITY, NORM_PRIORITY

Constructor Summary

AlgorithmEdgeNMSuppression(ModelImage destImg, ModelImage srcImg, float[] sigmas, boolean maskFlag, boolean img25D)
Creates a new AlgorithmEdgeNMSuppression object.

Method Summary

void	algorithmPerformed(AlgorithmBase algorithm) Called after an algorithm this listener is registered to exits (maybe successfully, maybe not).
void	finalize() Prepares this class for destruction.
static BitSet	genLevelMask(int xDim, int yDim, float[] buffer, float level, int detectionType) Generates a zero crossing mask for a 2D function sets a Bitset object to 1 is a zero crossing is detected.
void	genZeroXMask(int slice, float[] buffer, float[] buffer2, int detectionType) Generates a zero crossing mask for a 2D function sets a ModelImage to 255 if a zero crossing is detected.
ModelImage	getZeroXMask() Accessor to return mask indicating zero crossings.
void	runAlgorithm() Starts the program.

Methods inherited from class gov.nih.mipav.model.algorithms.AlgorithmBase

actionPerformed, addListener, addProgressChangeListener, calculatePrincipleAxis, computeElapsedTime, convertIntoFloat, delinkProgressToAlgorithm, displayError, errorCleanUp, fireProgressStateChanged, fireProgressStateChanged, fireProgressStateChanged, fireProgressStateChanged, fireProgressStateChanged, generateProgressValues, getDestImage, getElapsedTime, getMask, getMaxProgressValue, getMinProgressValue, getNumberOfThreads, getProgress, getProgressChangeListener, getProgressChangeListeners, getProgressModulus, getProgressStep, getProgressValues, isCompleted, isImage25D, isMultiThreadingEnabled, isRunningInSeparateThread, isThreadStopped, linkProgressToAlgorithm, makeProgress, notifyListeners, removeListener, removeProgressChangeListener, run, setCompleted, setImage25D, setMask, setMaxProgressValue, setMinProgressValue, setMultiThreadingEnabled, setNumberOfThreads, setProgress, setProgressModulus, setProgressStep, setProgressValues, setProgressValues, setRunningInSeparateThread, setStartTime, setThreadStopped, startMethod, windowActivated, windowClosed, windowClosing, windowDeactivated, windowDeiconified, windowIconified, windowOpened

Methods inherited from class java.lang.Thread

activeCount, checkAccess, countStackFrames, currentThread, destroy, dumpStack, enumerate, getAllStackTraces, getContextClassLoader, getDefaultUncaughtExceptionHandler, getId, getName, getPriority, getStackTrace, getState, getThreadGroup, getUncaughtExceptionHandler, holdsLock, interrupt, interrupted, isAlive, isDaemon, isInterrupted, join, join, join, resume, setContextClassLoader, setDaemon, setDefaultUncaughtExceptionHandler, setName, setPriority, setUncaughtExceptionHandler, sleep, sleep, start, stop, stop, suspend, toString, yield

Methods inherited from class java.lang.Object

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

Field Detail

MARCHING_SQUARES

public static final int MARCHING_SQUARES

Perform zero crossing detection using the marching squares method.

See Also:

Constant Field Values

OLD_DETECTION

public static final int OLD_DETECTION

Perform zero crossing detection using Matt's old method.

See Also:

Constant Field Values

Constructor Detail

AlgorithmEdgeNMSuppression

public AlgorithmEdgeNMSuppression(ModelImage destImg,
                                  ModelImage srcImg,
                                  float[] sigmas,
                                  boolean maskFlag,
                                  boolean img25D)

Creates a new AlgorithmEdgeNMSuppression object.

Parameters:

destImg - image model where result image is to stored

srcImg - source image model

sigmas - Gaussian's standard deviations in the each dimension

maskFlag - Flag that indicates that the EdgeNMSup will be calculated for the whole image if equal to true

img25D - Flag, if true, indicates that each slice of the 3D volume should be processed independently. 2D images disregard this flag.

Method Detail

genLevelMask

public static BitSet genLevelMask(int xDim,
                                  int yDim,
                                  float[] buffer,
                                  float level,
                                  int detectionType)

Generates a zero crossing mask for a 2D function sets a Bitset object to 1 is a zero crossing is detected.

Parameters:

xDim - the buffer's x dimension

yDim - the buffer's y dimension

buffer - array in which to find zero crossing

level - the level to find the crossing of (usually will be 0)

detectionType - the type of zero crossing method to use

Returns:

DOCUMENT ME!

finalize

public void finalize()

Prepares this class for destruction.

Overrides:

finalize in class AlgorithmBase

genZeroXMask

public void genZeroXMask(int slice,
                         float[] buffer,
                         float[] buffer2,
                         int detectionType)

Generates a zero crossing mask for a 2D function sets a ModelImage to 255 if a zero crossing is detected.

Parameters:

slice - DOCUMENT ME!

buffer - array in which to find zero crossing

buffer2 - array which ensures that zero crossing is only counted if buffer2 value at that position is less than zero

detectionType - the type of zero crossing detection to perform

getZeroXMask

public ModelImage getZeroXMask()

Accessor to return mask indicating zero crossings.

Returns:

ModelImage of zero crossings (2D function); 255 = indicates zero crossing

runAlgorithm

public void runAlgorithm()

Starts the program.

Specified by:

runAlgorithm in class AlgorithmBase

algorithmPerformed

public void algorithmPerformed(AlgorithmBase algorithm)

Description copied from interface: AlgorithmInterface

Called after an algorithm this listener is registered to exits (maybe successfully, maybe not). If the algorithm is run in a separate thread, this call will be made within that thread. If not, this call will be made from that same, shared thread.

Specified by:

algorithmPerformed in interface AlgorithmInterface

Parameters:

algorithm - the algorithm which has just completed