Package gov.nih.mipav.model.algorithms

Class AlgorithmMaximumLikelihoodIteratedBlindDeconvolution

java.lang.Object

java.lang.Thread

gov.nih.mipav.model.algorithms.AlgorithmBase

gov.nih.mipav.model.algorithms.AlgorithmMaximumLikelihoodIteratedBlindDeconvolution

All Implemented Interfaces:

ActionListener, WindowListener, Runnable, EventListener

public class AlgorithmMaximumLikelihoodIteratedBlindDeconvolution
extends AlgorithmBase

An implementation of Maximum Likelihood Iterated Blind Deconvolution based on the following papers:

Holmes, T.J., Maximum Likelihood Image Restoration Adapted for Noncoherent Optical Imaging, JOSA-A, 5(5): 666-673, 1988.

Holmes, T., Bhattacharyya, S., Cooper, J., Hanzel, D., Krishnamurthi, V., Lin, W., Roysam, B., Szarowski, D., Turner, J., Light Microscopic Images Reconstructed by Maximum Likelihood Deconvolution, Ch. 24, Handbook of Biological Confocal Microscopy, J. Pawley, Plenum, 1995.

Holmes, T., Liu, Y., Image Restoration for 2D and 3D Fluorescence Microscopy, Visualization in Biomedical Microscopies, A. Kriete, VCH, 1992.

Holmes, T., Blind Deconvolution of Quantum-Limited Noncoherent Imagery, JOSA-A, 9: 1052 - 1061, 1992.

Nested Class Summary

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

Thread.Builder, Thread.State, Thread.UncaughtExceptionHandler

Field Summary

Fields

Modifier and Type	Field	Description
private float[]	m_afResolutions	DOCUMENT ME!
private int[]	m_aiUnits	DOCUMENT ME!
private boolean	m_bConstraintsInit	DOCUMENT ME!
private boolean	m_bUpdatePBar	Flag indicating whether or not to update the progres bar:.
private boolean	m_bUseMicroscopeConstraints	Boolean to use constraints:.
private float	m_fAxialBandLimit	DOCUMENT ME!
private float[]	m_fNewRes	The new resolutions in each dimension if resample is true.
private float	m_fObjectiveNumericalAperature	numerical aperature of the imaging lens:.
private float	m_fRadialBandLimit	3).
private float	m_fRadius	DOCUMENT ME!
private float	m_fRefractiveIndex	sample index of refraction:.
private float	m_fTanTheta	DOCUMENT ME!
private float	m_fWavelength	wavelength of the reflected or fluorescing light:.
private int	m_iArrayLength	Original data size:.
private int	m_iDimX	Constraint parameters:.
private int	m_iDimY	DOCUMENT ME!
private int	m_iDimZ	DOCUMENT ME!
private int	m_iNumberIterations	Number of iterations for the deconvolution:.
private int	m_iNumberProgress	Show the deconvolved image in progress every m_iNumberProgress steps:.
private ModelImage	m_kCalcResult1	DOCUMENT ME!
private ModelImage	m_kCalcResult2	DOCUMENT ME!
private ModelImage	m_kCalcResult3	DOCUMENT ME!
private ModelImage	m_kEstimatedImage	estimated of the reconstructed image:.
private ModelImage	m_kImageSpectrum1	Images to store FFT and Calc results.
private ModelImage	m_kImageSpectrum2	DOCUMENT ME!
private ModelImage	m_kImageSpectrum3	DOCUMENT ME!
private ModelImage	m_kMirrorImage	DOCUMENT ME!
private ModelImage	m_kOriginalSourceImage	source image to be reconstructed!
private ModelImage	m_kPSFImage	point spread function image:.
private ModelImage	m_kPSFImageCopy	Used for color image.
private ModelImage	m_kSourceBlue	DOCUMENT ME!
private ModelImage	m_kSourceGreen	DOCUMENT ME!
private ModelImage	m_kSourceImage	source image to be reconstructed!
private ModelImage	m_kSourceRed	For color images: the source is converted to gray for the deconvolution.

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

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

Fields inherited from class java.lang.Thread

MAX_PRIORITY, MIN_PRIORITY, NORM_PRIORITY

Constructor Summary

Constructors

Constructor	Description
AlgorithmMaximumLikelihoodIteratedBlindDeconvolution(ModelImage kSrcImg, int iIterations, int iProgress, float fObjectiveNumericalAperature, float fWavelength, float fRefractiveIndex, boolean bUseConstraints, int[] newExtents, boolean doResample)	Creates a new AlgorithmMaximumLikelihoodIteratedBlindDeconvolution object.

Method Summary

Modifier and Type	Method	Description
private ModelImage	calc(ModelImage kReturn, ModelImage kImage1, ModelImage kImage2, int iType)	Performs a AlgorithmImageCalculator calculation on the two input images.
private void	cleanUp()	Dispose of temporary images after the deconvolution process is completed:
private ModelImage	constraints(ModelImage kImagePSF)	Applies constraints to the input image kImage and writes the constrained values into the m_kPSFImage.
private ModelImage	convertFromGray(ModelImage kRed, ModelImage kGreen, ModelImage kBlue)	Called after the deconvolution algorithm is run.
private ModelImage	convertToGray(ModelImage kImage)	Convert the input image, kImage, to a grayscale image using the AlgorithmRGBtoGray.
private ModelImage	convolve(ModelImage kImage1, ModelImage kImage2, float fNoise)	Convolves two image by computing the fast fourier transforms of each image, multiplying the ffts and then computing the inverse fft of the result:
void	disposeLocal()	Dispose of local variables that may be taking up lots of room.
private ModelImage	fft(ModelImage kImageFFT, ModelImage kImage, int iDir)	Computes the fft of the input image, returns the fft.
void	finalize()	Prepares this class for destruction.
ModelImage	getPSFImage()	Returns the point spread function image.
ModelImage	getReconstructedImage()	Returns the reconstructed image:
private void	initConstraints()	Initializes the constraint variables for the PSF deconvolution, based on the microscope parameters (numerical aperature, wavelength, and refractive index).
private void	initPSF(int iNumberDimensions, int[] aiExtents)	Initialize the first PSF guess as a 3x3 Gaussian.
private ModelImage	mirror(ModelImage kImage)	Does a mirror operation on an image, so that x, y, and z values are mirrored Calls JDialogFlip.
private ModelImage	resample(ModelImage kImage, int[] newExtents)
void	runAlgorithm()	Starts the program.
private ModelImage	runDeconvolution(ModelImage kSource, ModelImage kEstimate, ModelImage kPSF, boolean bCopy)	Runs the deconvolution algorithm.

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

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

Methods inherited from class java.lang.Thread

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

Methods inherited from class java.lang.Object

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

Field Details

m_afResolutions

private float[] m_afResolutions

DOCUMENT ME!

m_aiUnits

private int[] m_aiUnits

DOCUMENT ME!

m_bConstraintsInit

private boolean m_bConstraintsInit

DOCUMENT ME!

m_bUpdatePBar

private boolean m_bUpdatePBar

Flag indicating whether or not to update the progres bar:.

m_bUseMicroscopeConstraints

private boolean m_bUseMicroscopeConstraints

Boolean to use constraints:.

m_fAxialBandLimit

private float m_fAxialBandLimit

DOCUMENT ME!

m_fObjectiveNumericalAperature

private float m_fObjectiveNumericalAperature

numerical aperature of the imaging lens:.

m_fRadialBandLimit

private float m_fRadialBandLimit

3). Bandlimit and missing cone constraint:

m_fRadius

private float m_fRadius

DOCUMENT ME!

m_fRefractiveIndex

private float m_fRefractiveIndex

sample index of refraction:.

m_fTanTheta

private float m_fTanTheta

DOCUMENT ME!

m_fWavelength

private float m_fWavelength

wavelength of the reflected or fluorescing light:.

m_iArrayLength

private int m_iArrayLength

Original data size:.

m_iDimX

private int m_iDimX

Constraint parameters:.

m_iDimY

private int m_iDimY

DOCUMENT ME!

m_iDimZ

private int m_iDimZ

DOCUMENT ME!

m_iNumberIterations

private int m_iNumberIterations

Number of iterations for the deconvolution:.

m_iNumberProgress

private int m_iNumberProgress

Show the deconvolved image in progress every m_iNumberProgress steps:.

m_fNewRes

private float[] m_fNewRes

The new resolutions in each dimension if resample is true. Used only if resample is true.

m_kCalcResult1

private ModelImage m_kCalcResult1

DOCUMENT ME!

m_kCalcResult2

private ModelImage m_kCalcResult2

DOCUMENT ME!

m_kCalcResult3

private ModelImage m_kCalcResult3

DOCUMENT ME!

m_kEstimatedImage

private ModelImage m_kEstimatedImage

estimated of the reconstructed image:.

m_kImageSpectrum1

private ModelImage m_kImageSpectrum1

Images to store FFT and Calc results.

m_kImageSpectrum2

private ModelImage m_kImageSpectrum2

DOCUMENT ME!

m_kImageSpectrum3

private ModelImage m_kImageSpectrum3

DOCUMENT ME!

m_kMirrorImage

private ModelImage m_kMirrorImage

DOCUMENT ME!

m_kOriginalSourceImage

private ModelImage m_kOriginalSourceImage

source image to be reconstructed!

m_kPSFImage

private ModelImage m_kPSFImage

point spread function image:.

m_kPSFImageCopy

private ModelImage m_kPSFImageCopy

Used for color image.

m_kSourceBlue

private ModelImage m_kSourceBlue

DOCUMENT ME!

m_kSourceGreen

private ModelImage m_kSourceGreen

DOCUMENT ME!

m_kSourceImage

private ModelImage m_kSourceImage

source image to be reconstructed!

m_kSourceRed

private ModelImage m_kSourceRed

For color images: the source is converted to gray for the deconvolution. Once the estimate and psf are found, each component is reconstructed.

Constructor Details

AlgorithmMaximumLikelihoodIteratedBlindDeconvolution

public AlgorithmMaximumLikelihoodIteratedBlindDeconvolution(ModelImage kSrcImg,
 int iIterations,
 int iProgress,
 float fObjectiveNumericalAperature,
 float fWavelength,
 float fRefractiveIndex,
 boolean bUseConstraints,
 int[] newExtents,
 boolean doResample)

Creates a new AlgorithmMaximumLikelihoodIteratedBlindDeconvolution object.

Parameters:

kSrcImg - the input image to be reconstructed

iIterations - the number of times to iterate in the deconvolution

iProgress - Display deconvolved image every iProgress images

fObjectiveNumericalAperature - the numerical aperature of the imaging lense

fWavelength - the reflected or fluorescing light

fRefractiveIndex - the index of refraction for the sample

bUseConstraints - When true, the lens NA, wavelength, and refractive index are used in the deconvolution process

Method Details

disposeLocal

public void disposeLocal()

Dispose of local variables that may be taking up lots of room.

finalize

public void finalize()

Prepares this class for destruction.

Overrides:

finalize in class AlgorithmBase

getPSFImage

public ModelImage getPSFImage()

Returns the point spread function image.

Returns:

m_kPSFImage

getReconstructedImage

public ModelImage getReconstructedImage()

Returns the reconstructed image:

Returns:

m_kEstimatedImage

runAlgorithm

public void runAlgorithm()

Starts the program.

Specified by:

runAlgorithm in class AlgorithmBase

calc

private ModelImage calc(ModelImage kReturn,
 ModelImage kImage1,
 ModelImage kImage2,
 int iType)

Performs a AlgorithmImageCalculator calculation on the two input images. Used here to either multiply or divide two images.

Parameters:

kReturn - DOCUMENT ME!

kImage1 - the first input image

kImage2 - the second input image

iType - the type of calculation (multiply or divide)

Returns:

kReturn, the result of the image calculation

cleanUp

private void cleanUp()

Dispose of temporary images after the deconvolution process is completed:

constraints

private ModelImage constraints(ModelImage kImagePSF)

Applies constraints to the input image kImage and writes the constrained values into the m_kPSFImage. Constraints are the following: 2). Hourglass constraint 3). Bandlimit and missing cone constraint 4). Nonnegativity constraint 1). Unit summation constraint

Parameters:

kImagePSF - the unconstrained PSFImage

Returns:

DOCUMENT ME!

convertFromGray

private ModelImage convertFromGray(ModelImage kRed,
 ModelImage kGreen,
 ModelImage kBlue)

Called after the deconvolution algorithm is run. The deconvolution processes the grayscale image. Once the estimated and psf images are calculated, they are used to reconstruct the red, green, and blue channels separately, which are then recombined into a color image.

Parameters:

kRed - the reconstructed red image

kGreen - the reconstructed green image

kBlue - the reconstructed blue image

Returns:

the RGB image

convertToGray

private ModelImage convertToGray(ModelImage kImage)

Convert the input image, kImage, to a grayscale image using the AlgorithmRGBtoGray. The deconvolution is done on the grayscale image, however after deconvolution, the original color values are reconstructed from the psf and grayscale estimate.

Parameters:

kImage - the color image to conver to grayscale:

Returns:

the converted grayscale image

convolve

private ModelImage convolve(ModelImage kImage1,
 ModelImage kImage2,
 float fNoise)

Convolves two image by computing the fast fourier transforms of each image, multiplying the ffts and then computing the inverse fft of the result:

Parameters:

kImage1 - the first input image

kImage2 - the second input image

fNoise - amount of Noise to add when the result of the convolution is going to be the denominator in the next equation to avoid divide by zero errors.

Returns:

kResult, the result of convolving the two images

fft

private ModelImage fft(ModelImage kImageFFT,
 ModelImage kImage,
 int iDir)

Computes the fft of the input image, returns the fft. Calls AlgorithmFFT.

Parameters:

kImageFFT - DOCUMENT ME!

kImage - the input image

iDir - forward fft when iDir = 1, inverse fft when iDir = -1

Returns:

kResult, the fft of kImage

initConstraints

private void initConstraints()

Initializes the constraint variables for the PSF deconvolution, based on the microscope parameters (numerical aperature, wavelength, and refractive index). The pixel resolutions are also converted to nanometers -- the units for wavelength.

initPSF

private void initPSF(int iNumberDimensions,
 int[] aiExtents)

Initialize the first PSF guess as a 3x3 Gaussian.

Parameters:

iNumberDimensions - iNumberDimensions, the dimensions of the data

aiExtents - aiExtents, the ranges in each dimension

mirror

private ModelImage mirror(ModelImage kImage)

Does a mirror operation on an image, so that x, y, and z values are mirrored Calls JDialogFlip.

Parameters:

kImage - the input image

Returns:

DOCUMENT ME!

resample

private ModelImage resample(ModelImage kImage,
 int[] newExtents)

runDeconvolution

private ModelImage runDeconvolution(ModelImage kSource,
 ModelImage kEstimate,
 ModelImage kPSF,
 boolean bCopy)

Runs the deconvolution algorithm.

Parameters:

kSource - the original image data

kEstimate - the initial estimated guess at the reconstructed image

kPSF - the initial point spread function guess

bCopy - when true the new estimates after interating are copied back into the data members.

Returns:

, the resulting estimated image