def get(self, index):
img = CellLoader.klb.readFull(timepoint_paths[index]).getImg()
# Each cell has "1" as its dimension in the last axis (time)
# and index as its min coordinate in the last axis (time)
return Cell(
Intervals.dimensionsAsIntArray(img) + array([1], 'i'),
Intervals.minAsLongArray(img) + array([index], 'l'),
extractArrayAccess(img))
def makeInterpolatedImage(img1, img2, weight):
""" weight: float between 0 and 1 """
edge_pix1 = findEdgePixels(img1)
kdtree1 = KDTree(edge_pix1, edge_pix1)
search1 = NearestNeighborSearchOnKDTree(kdtree1)
edge_pix2 = findEdgePixels(img2)
kdtree2 = KDTree(edge_pix2, edge_pix2)
search2 = NearestNeighborSearchOnKDTree(kdtree2)
img3 = ArrayImgs.unsignedBytes(Intervals.dimensionsAsLongArray(img1))
c1 = img1.cursor()
c2 = img2.cursor()
c3 = img3.cursor()
pos = zeros(img1.numDimensions(), 'l')
while c3.hasNext():
t1 = c1.next()
t2 = c2.next()
t3 = c3.next()
sign1 = -1 if 0 == t1.get() else 1
sign2 = -1 if 0 == t2.get() else 1
search1.search(c1)
search2.search(c2)
value1 = sign1 * search1.getDistance() * weight
value2 = sign2 * search2.getDistance() * (1 - weight)
if value1 + value2 > 0:
t3.setOne()
return img3
def asNormalizedUnsignedByteArrayImg(interval, invert, blockRadius, n_bins,
slope, matrices, copy_threads, index,
imp):
sp = imp.getProcessor() # ShortProcessor
sp.setRoi(interval.min(0), interval.min(1),
interval.max(0) - interval.min(0) + 1,
interval.max(1) - interval.min(1) + 1)
sp = sp.crop()
if invert:
sp.invert()
CLAHE.run(
ImagePlus("", sp), blockRadius, n_bins, slope, None
) # far less memory requirements than NormalizeLocalContrast, and faster.
minimum, maximum = autoAdjust(sp)
# Transform and convert image to 8-bit, mapping to display range
img = ArrayImgs.unsignedShorts(
sp.getPixels(), [sp.getWidth(), sp.getHeight()])
sp = None
affine = AffineTransform2D()
affine.set(matrices[index])
imgI = Views.interpolate(Views.extendZero(img),
NLinearInterpolatorFactory())
imgA = RealViews.transform(imgI, affine)
imgT = Views.zeroMin(Views.interval(imgA, img))
imgMinMax = convert(imgT, RealUnsignedByteConverter(minimum, maximum),
UnsignedByteType)
aimg = ArrayImgs.unsignedBytes(Intervals.dimensionsAsLongArray(img))
ImgUtil.copy(ImgView.wrap(imgMinMax, aimg.factory()), aimg,
copy_threads)
img = imgI = imgA = imgT = imgMinMax = None
return aimg
def crop_along_one_axis(ops, data, intervals, axis_type):
"""Crop along a single axis using Views.
Parameters
----------
intervals : List with two values specifying the start and the end of the interval.
axis_type : Along which axis to crop. Can be ["X", "Y", "Z", "TIME", "CHANNEL"]
"""
axis = get_axis(axis_type)
interval_start = [
data.min(d) if d != data.dimensionIndex(axis) else intervals[0]
for d in range(0, data.numDimensions())
]
interval_end = [
data.max(d) if d != data.dimensionIndex(axis) else intervals[1]
for d in range(0, data.numDimensions())
]
interval = interval_start + interval_end
interval = Intervals.createMinMax(*interval)
output = ops.run("transform.crop", data, interval, True)
return output
def test(iraf):
# Test dimensions: should be the same as the one input image
print "Dimensions:", Intervals.dimensionsAsLongArray(iraf)
# Test Cursor
c = iraf.cursor()
pos = zeros(2, 'l')
while c.hasNext():
c.fwd()
c.localize(pos)
print "Cursor:", pos, "::", c.get()
# Test RandomAccess
ra = iraf.randomAccess()
c = iraf.cursor()
while c.hasNext():
c.fwd()
ra.setPosition(c)
c.localize(pos)
print "RandomAccess:", pos, "::", ra.get()
# Test source img: should be untouched
c = img.cursor()
while c.hasNext():
print "source:", c.next()
# Test interval view: the middle 2x2 square
v = Views.interval(iraf, [1, 1], [2, 2])
IL.wrap(v, "+2 view").show()
def rotatedView(img, angle, enlarge=True, extend=Views.extendBorder):
""" Return a rotated view of the image, around the Z axis,
with an expanded (or reduced) interval view so that all pixels are exactly included.
img: a RandomAccessibleInterval
angle: in degrees
"""
cx = img.dimension(0) / 2.0
cy = img.dimension(1) / 2.0
toCenter = AffineTransform2D()
toCenter.translate(-cx, -cy)
rotation = AffineTransform2D()
# Step 1: place origin of rotation at the center of the image
rotation.preConcatenate(toCenter)
# Step 2: rotate around the Z axis
rotation.rotate(radians(angle))
# Step 3: undo translation to the center
rotation.preConcatenate(toCenter.inverse())
rotated = RV.transform(Views.interpolate(extend(img),
NLinearInterpolatorFactory()), rotation)
if enlarge:
# Bounds:
bounds = repeat((sys.maxint, 0)) # initial upper- and lower-bound values
# for min, max to compare against
transformed = zeros(2, 'f')
for corner in product(*zip(repeat(0), Intervals.maxAsLongArray(img))):
rotation.apply(corner, transformed)
bounds = [(min(vmin, int(floor(v))), max(vmax, int(ceil(v))))
for (vmin, vmax), v in zip(bounds, transformed)]
minC, maxC = map(list, zip(*bounds)) # transpose list of 2 pairs
# into 2 lists of 2 values
imgRot = Views.zeroMin(Views.interval(rotated, minC, maxC))
else:
imgRot = Views.interval(rotated, img)
return imgRot
def twoStep(index=0):
# The current way:
img = klb.readFull(filepaths[index]) # klb_loader.get(filepaths[index])
imgE = Views.extendZero(img)
imgI = Views.interpolate(imgE, NLinearInterpolatorFactory())
imgT = RealViews.transform(imgI, cmIsotropicTransforms[index])
imgB = Views.zeroMin(Views.interval(imgT, roi[0],
roi[1])) # bounded: crop with ROI
imgBA = ArrayImgs.unsignedShorts(Intervals.dimensionsAsLongArray(imgB))
ImgUtil.copy(ImgView.wrap(imgB, imgBA.factory()), imgBA)
imgP = prepareImgForDeconvolution(
imgBA,
affine3D(fineTransformsPostROICrop[index]).inverse(),
FinalInterval([0, 0, 0], [imgB.dimension(d) - 1 for d in xrange(3)]))
# Copy transformed view into ArrayImg for best performance in deconvolution
imgA = ArrayImgs.floats(Intervals.dimensionsAsLongArray(imgP))
ImgUtil.copy(ImgView.wrap(imgP, imgA.factory()), imgA)
IL.wrap(imgA, "two step").show()
def translate(self, dx, dy):
a = zeros(2, 'l')
self.interval.min(a)
width = self.cell_dimensions[0]
height = self.cell_dimensions[1]
x0 = max(0, min(a[0] + dx, self.img_dimensions[0] - width))
y0 = max(0, min(a[1] + dy, self.img_dimensions[1] - height))
self.interval = FinalInterval([x0, y0],
[x0 + width - 1, y0 + height - 1])
syncPrintQ(str(Intervals.dimensionsAsLongArray(self.interval)))
self.cache.clear()
def projectMax(img, minC, maxC, reduce_max):
imgA = ArrayImgs.unsignedSorts(
Intervals.dimensionsAsLongArray(imgC))
ImgUtil.copy(
ImgView.wrap(
convert(
Views.collapseReal(
Views.interval(img, minC, maxC)),
reduce_max.newInstance(), imglibtype),
img.factory()), imgA)
return imgA
def makeImg(filepaths, pixelType, loadImg, img_dimensions, matrices,
cropInterval, preload):
dims = Intervals.dimensionsAsLongArray(cropInterval)
voldims = [dims[0], dims[1], len(filepaths)]
cell_dimensions = [dims[0], dims[1], 1]
grid = CellGrid(voldims, cell_dimensions)
cellGet = TranslatedSectionGet(filepaths,
loadImg,
matrices,
img_dimensions,
cell_dimensions,
cropInterval,
preload=preload)
return LazyCellImg(grid, pixelType(), cellGet), cellGet
def crop(event):
global cropped, cropped_imp
coords = [int(float(tf.getText())) for tf in textfields]
minC = [max(0, c) for c in coords[0:3]]
maxC = [min(d -1, c) for d, c in izip(Intervals.dimensionsAsLongArray(images[0]), coords[3:6])]
storeRoi(minC, maxC)
print "ROI min and max coordinates"
print minC
print maxC
cropped = [Views.zeroMin(Views.interval(img, minC, maxC)) for img in images]
cropped_imp = showAsStack(cropped, title="cropped")
cropped_imp.setDisplayRange(imp.getDisplayRangeMin(), imp.getDisplayRangeMax())
if cropContinuationFn:
cropContinuationFn(images, minC, maxC, cropped, cropped_imp)
def populateInstances(instances, synth_imgs, class_index, mins, maxs):
# Populate the training data: create the filter bank for each feature image
# by reading values from the interval defined by mins and maxs
target = ArrayImgs.floats([width, height])
interval = FinalInterval(mins, maxs)
n_samples = Intervals.numElements(interval)
for img in synth_imgs:
vectors = [zeros(len(attributes), 'd') for _ in xrange(n_samples)]
for k, op in enumerate(filterBank(img, sumType=DoubleType())):
imgOp = compute(op).into(target)
for i, v in enumerate(Views.interval(imgOp, interval)):
vectors[i][k] = v.getRealDouble()
for vector in vectors:
vector[-1] = class_index
instances.add(DenseInstance(1.0, vector))
def asNormalizedUnsignedByteArrayImg(interval, invert, blockRadius, n_bins,
slope, matrices, index, imp):
sp = imp.getProcessor() # ShortProcessor
# Crop to interval if needed
x = interval.min(0)
y = interval.min(1)
width = interval.max(0) - interval.min(0) + 1
height = interval.max(1) - interval.min(1) + 1
if 0 != x or 0 != y or sp.getWidth() != width or sp.getHeight(
) != height:
sp.setRoi(x, y, width, height)
sp = sp.crop()
if invert:
sp.invert()
CLAHE.run(
ImagePlus("", sp), blockRadius, n_bins, slope, None
) # far less memory requirements than NormalizeLocalContrast, and faster.
minimum, maximum = autoAdjust(sp)
# Transform and convert image to 8-bit, mapping to display range
img = ArrayImgs.unsignedShorts(
sp.getPixels(), [sp.getWidth(), sp.getHeight()])
sp = None
imp = None
# Must use linear interpolation for subpixel precision
affine = AffineTransform2D()
affine.set(matrices[index])
imgI = Views.interpolate(Views.extendZero(img),
NLinearInterpolatorFactory())
imgA = RealViews.transform(imgI, affine)
imgT = Views.zeroMin(Views.interval(imgA, img))
# Convert to 8-bit
imgMinMax = convert2(imgT,
RealUnsignedByteConverter(minimum, maximum),
UnsignedByteType,
randomAccessible=False) # use IterableInterval
aimg = ArrayImgs.unsignedBytes(Intervals.dimensionsAsLongArray(img))
# ImgUtil copies multi-threaded, which is not appropriate here as there are many other images being copied too
#ImgUtil.copy(ImgView.wrap(imgMinMax, aimg.factory()), aimg)
# Single-threaded copy
copier = createBiConsumerTypeSet(UnsignedByteType)
LoopBuilder.setImages(imgMinMax, aimg).forEachPixel(copier)
img = imgI = imgA = imgMinMax = imgT = None
return aimg
def prepare(index):
# Prepare the img for deconvolution:
# 0. Transform in one step.
# 1. Ensure its pixel values conform to expectations (no zeros inside)
# 2. Copy it into an ArrayImg for faster recurrent retrieval of same pixels
syncPrint("Preparing %s CM0%i for deconvolution" % (tm_dirname, index))
img = klb_loader.get(filepaths[index]) # of UnsignedShortType
imgP = prepareImgForDeconvolution(
img, transforms[index], target_interval) # returns of FloatType
# Copy transformed view into ArrayImg for best performance in deconvolution
imgA = ArrayImgs.floats(Intervals.dimensionsAsLongArray(imgP))
#ImgUtil.copy(ImgView.wrap(imgP, imgA.factory()), imgA)
ImgUtil.copy(imgP, imgA, n_threads / 2) # parallel copying
syncPrint("--Completed preparing %s CM0%i for deconvolution" %
(tm_dirname, index))
imgP = None
img = None
return (index, imgA)
def crop_along_one_axis(ops, data, intervals, axis_type):
"""Crop along a single axis using Views.
Parameters
----------
intervals : List with two values specifying the start and the end of the interval.
axis_type : Along which axis to crop. Can be ["X", "Y", "Z", "TIME", "CHANNEL"]
"""
axis = get_axis(axis_type)
interval_start = [data.min(d) if d != data.dimensionIndex(axis) else intervals[0] for d in range(0, data.numDimensions())]
interval_end = [data.max(d) if d != data.dimensionIndex(axis) else intervals[1] for d in range(0, data.numDimensions())]
interval = interval_start + interval_end
interval = Intervals.createMinMax(*interval)
output = ops.run("transform.crop", data, interval, True)
return output
def makeInterpolatedImage(img1, search1, img2, search2, weight):
""" weight: float between 0 and 1 """
img3 = ArrayImgs.unsignedBytes(Intervals.dimensionsAsLongArray(img1))
c1 = img1.cursor()
c2 = img2.cursor()
c3 = img3.cursor()
while c3.hasNext():
t1 = c1.next()
t2 = c2.next()
t3 = c3.next()
sign1 = -1 if 0 == t1.get() else 1
sign2 = -1 if 0 == t2.get() else 1
search1.search(c1)
search2.search(c2)
value1 = sign1 * search1.getDistance() * (1 - weight)
value2 = sign2 * search2.getDistance() * weight
if value1 + value2 > 0:
t3.setOne()
return img3
def oneStep(index=0):
# Combining transforms into one, via a translation to account of the ROI crop
img = klb.readFull(filepaths[index]) # klb_loader.get(filepaths[index])
t1 = cmIsotropicTransforms[index]
t2 = affine3D(
[1, 0, 0, -roi[0][0], 0, 1, 0, -roi[0][1], 0, 0, 1, -roi[0][2]])
t3 = affine3D(fineTransformsPostROICrop[index]).inverse()
aff = AffineTransform3D()
aff.set(t1)
aff.preConcatenate(t2)
aff.preConcatenate(t3)
# Final interval is now rooted at 0,0,0 given that the transform includes the translation
imgP = prepareImgForDeconvolution(
img, aff,
FinalInterval([0, 0, 0],
[maxC - minC for minC, maxC in izip(roi[0], roi[1])]))
# Copy transformed view into ArrayImg for best performance in deconvolution
imgA = ArrayImgs.floats(Intervals.dimensionsAsLongArray(imgP))
ImgUtil.copy(ImgView.wrap(imgP, imgA.factory()), imgA)
IL.wrap(imgA, "one step index %i" % index).show()
def filterBankRotations(img,
angles=xrange(0, 46, 9), # sequence, in degrees
filterBankFn=filterBank, # function that takes an img as sole positional argument
outputType=FloatType()):
""" img: a RandomAccessibleInterval.
filterBankFn: the function from which to obtain a sequence of ImgMath ops.
angles: a sequence of angles in degrees.
outputType: for materializing rotated operations and rotating them back.
For every angle, will prepare a rotated view of the image,
then create a list of ops on the basis of that rotated view,
then materialize each op into an image so that an unrotated view
can be returned back.
returns a list of unrotated views, each containing the values of applying
each op to the rotated view.
"""
ops_rotations = []
for angle in angles:
imgRot = img if 0 == angle else rotatedView(img, angle)
ops = filterBankFn(imgRot)
# Materialize these two combination ops and rotate them back (rather, a rotated view)
interval = Intervals.translate(img, [(imgRot.dimension(d) - img.dimension(d)) / 2
for d in xrange(img.numDimensions())])
for op in ops:
imgOpRot = compute(op).intoArrayImg(outputType)
if 0 == angle:
ops_rotations.append(imgOpRot)
continue
# Rotate them back and crop view
imgOpUnrot = rotatedView(imgOpRot, -angle, enlarge=False)
imgOp = Views.zeroMin(Views.interval(imgOpUnrot, interval))
#if angle == 0 or angle == 45:
# IL.wrap(imgOpRot, "imgOpRot angle=%i" % angle).show()
# IL.wrap(imgOpUnrot, "imgOpUnrot angle=%i" % angle).show()
# IL.wrap(imgOp, "imgOp angle=%i" % angle).show()
ops_rotations.append(imgOp)
return ops_rotations
def crop(ops, data, intervals):
"""Crop along a one or more axis.
Parameters
----------
intervals : Dict specifying which axis to crop and with what intervals.
Example :
intervals = {'X' : [0, 50],
'Y' : [0, 50]}
"""
intervals_start = [data.min(d) for d in range(0, data.numDimensions())]
intervals_end = [data.max(d) for d in range(0, data.numDimensions())]
for axis_type, interval in intervals.items():
index = data.dimensionIndex(get_axis(axis_type))
intervals_start[index] = interval[0]
intervals_end[index] = interval[1]
intervals = Intervals.createMinMax(*intervals_start + intervals_end)
output = ops.run("transform.crop", data, intervals, True)
return output
def crop(ops, data, intervals):
"""Crop along a one or more axis.
Parameters
----------
intervals : Dict specifying which axis to crop and with what intervals.
Example :
intervals = {'X' : [0, 50],
'Y' : [0, 50]}
"""
intervals_start = [data.min(d) for d in range(0, data.numDimensions())]
intervals_end = [data.max(d) for d in range(0, data.numDimensions())]
for axis_type, interval in intervals.items():
index = data.dimensionIndex(get_axis(axis_type))
intervals_start[index] = interval[0]
intervals_end[index] = interval[1]
intervals = Intervals.createMinMax(*intervals_start + intervals_end)
output = ops.run("transform.crop", data, intervals, True)
return output
aff = AffineTransform3D()
aff.set(*matrix)
return aff
# Transform the kernel for each view
kernels = [kernel,
transformPSFKernelToView(kernel, affine3D(matrices["imgB0-imgB1"])),
transformPSFKernelToView(kernel, affine3D(matrices["imgB0-imgB2"])),
transformPSFKernelToView(kernel, affine3D(matrices["imgB0-imgB3"]))]
def deconvolve(images, kernels, name, n_iterations):
# Bayesian-based multi-view deconvolution
exe = newFixedThreadPool(Runtime.getRuntime().availableProcessors() -2)
try:
mylambda = 0.0006
blockSize = Intervals.dimensionsAsIntArray(images[0]) # [128, 128, 128]
cptf = ComputeBlockSeqThreadCPUFactory(exe, mylambda, blockSize, ArrayImgFactory(FloatType()))
psiInitFactory = PsiInitBlurredFusedFactory() # PsiInitAvgPreciseFactory() fails with type mismatch: UnsignedByteType (?) vs FloatType
weight = Views.interval(ConstantRandomAccessible(FloatType(1), images[0].numDimensions()), FinalInterval(images[0]))
filterBlocksForContent = False # Run once with True, none were removed
decon_views = DeconViews([DeconView(exe, img, weight, kernel, PSFTYPE.INDEPENDENT, blockSize, 1, filterBlocksForContent)
for img in images],
exe)
#n_iterations = 10
decon = MultiViewDeconvolutionSeq(decon_views, n_iterations, psiInitFactory, cptf, ArrayImgFactory(FloatType()))
if not decon.initWasSuccessful():
print "Something went wrong initializing MultiViewDeconvolution"
else:
decon.runIterations()
img = decon.getPSI()
imp = IL.wrap(img, name + "_deconvolved_" + str(n_iterations) + "_iterations")
print median, max_sum
# Turns out the maximum is infinity.
# Therefore, discard all infinity values, and also any above 1.5 * median
threshold = median * 1.5
filtered = [
filename for filename, pixel_sum in sums if pixel_sum < threshold
]
n_threads = Runtime.getRuntime().availableProcessors()
threads = []
chunk_size = len(filtered) / n_threads
aimgs = []
first = klb.readFull(os.path.join(srcDir, filtered[0]))
dimensions = Intervals.dimensionsAsLongArray(first)
for i in xrange(n_threads):
m = Max(dimensions, filtered[i * chunk_size:(i + 1) * chunk_size])
m.start()
threads.append(m)
# Await completion of all
for m in threads:
m.join()
# Merge all results into a single maximum projection
max_projection = computeInto(maximum([m.aimg for m in threads]),
ArrayImgs.floats(dimensions))
max3D = writeZip(max_projection,
from net.imglib2.util import Intervals
from net.imagej.axis import Axes
from net.imglib2.type.numeric.real import FloatType
from java.lang.Math import floor
# crop PSF to desired size, this makes decon run faster with little effect on final quality
psfX = psf.dimension(data.dimensionIndex(Axes.X))
psfY = psf.dimension(data.dimensionIndex(Axes.Y))
psfZ = psf.dimension(2)
psf_ = ops.transform().crop(
psf.getImgPlus(),
Intervals.createMinMax(psfX / 2 - psfXSize / 2, psfY / 2 - psfYSize / 2, 0,
psfX / 2 + psfXSize / 2 - 1,
psfY / 2 + psfYSize / 2 - 1, psfZ - 1))
psf_ = ops.convert().float32(psf_)
maxPSF = ops.stats().max(psf_).getRealFloat()
psfBackground = psfBackgroundPercent * maxPSF
# subtract background from psf
for t in psf_:
val = t.getRealFloat() - psfBackground
if val < 0:
val = 0
t.setReal(val)
# normalize psf
IL.wrap(img_sub, "LoopBuilder").show()
"""
# Example 2b: with ImgLib2 LoopBuilder using a clojure-defined TriConsumer
from net.imglib2.img.display.imagej import ImageJFunctions as IL
from net.imglib2.converter import Converters
from net.imglib2.img.array import ArrayImgs
from net.imglib2.util import Intervals
from net.imglib2.loops import LoopBuilder
from org.scijava.plugins.scripting.clojure import ClojureScriptEngine
img = IL.wrap(imp_rgb) # an ARGBType Img
red = Converters.argbChannel(img, 1) # a view of the ARGB red channel
green = Converters.argbChannel(img, 2) # a view of the ARGB green channel
img_sub = ArrayImgs.unsignedBytes(Intervals.dimensionsAsLongArray(img)) # the img to store the result
code = """
(deftype Consumer [^long threshold]
%s
(accept [self red green result] ; can't type-hint, doesn't find matching method
(let [^%s r red
^%s g green
^%s s result]
(.setInteger s (if (>= (.getInteger r) threshold)
(.getInteger g)
0)))))
""" % ((LoopBuilder.TriConsumer.getName(),) \
+ tuple(a.randomAccess().get().getClass().getName() for a in [red, green, img_sub]))
clj = ClojureScriptEngine()
def makeCropUI(imp, images, tgtDir, panel=None, cropContinuationFn=None):
""" imp: the ImagePlus to work on.
images: the list of ImgLib2 images, one per frame, not original but already isotropic.
(These are views that use a nearest neighbor interpolation using the calibration to scale to isotropy.)
tgtDir: the target directory where e.g. CSV files will be stored, for ROI, features, pointmatches.
panel: optional, a JPanel controlled by a GridBagLayout.
cropContinuationFn: optional, a function to execute after cropping,
which is given as arguments the original images,
minC, maxC (both define a ROI), and the cropped images. """
independent = None == panel
if not panel:
panel = JPanel()
panel.setBorder(BorderFactory.createEmptyBorder(10,10,10,10))
gb = GridBagLayout()
gc = GBC()
else:
gb = panel.getLayout()
# Constraints of the last component
gc = gb.getConstraints(panel.getComponent(panel.getComponentCount() - 1))
# Horizontal line to separate prior UI components from crop UI
gc.gridx = 0
gc.gridy += 1
gc.gridwidth = 4
gc.anchor = GBC.WEST
gc.fill = GBC.HORIZONTAL
sep = JSeparator()
sep.setMinimumSize(Dimension(200, 10))
gb.setConstraints(sep, gc)
panel.add(sep)
# ROI UI header
title = JLabel("ROI controls:")
gc.gridy +=1
gc.anchor = GBC.WEST
gc.gridwidth = 4
gb.setConstraints(title, gc)
panel.add(title)
# Column labels for the min and max coordinates
gc.gridy += 1
gc.gridwidth = 1
for i, title in enumerate(["", "X", "Y", "Z"]):
gc.gridx = i
gc.anchor = GBC.CENTER
label = JLabel(title)
gb.setConstraints(label, gc)
panel.add(label)
textfields = []
rms = []
# Load stored ROI if any
roi_path = path = os.path.join(tgtDir, "crop-roi.csv")
if os.path.exists(roi_path):
with open(roi_path, 'r') as csvfile:
reader = csv.reader(csvfile, delimiter=',', quotechar="\"")
reader.next() # header
minC = map(int, reader.next()[1:])
maxC = map(int, reader.next()[1:])
# Place the ROI over the ImagePlus
imp.setRoi(Roi(minC[0], minC[1], maxC[0] + 1 - minC[0], maxC[1] + 1 - minC[1]))
else:
# Use whole image dimensions
minC = [0, 0, 0]
maxC = [v -1 for v in Intervals.dimensionsAsLongArray(images[0])]
# Text fields for the min and max coordinates
for rowLabel, coords in izip(["min coords: ", "max coords: "],
[minC, maxC]):
gc.gridx = 0
gc.gridy += 1
label = JLabel(rowLabel)
gb.setConstraints(label, gc)
panel.add(label)
for i in xrange(3):
gc.gridx += 1
tf = JTextField(str(coords[i]), 10)
gb.setConstraints(tf, gc)
panel.add(tf)
textfields.append(tf)
listener = RoiMaker(imp, textfields, len(textfields) -1)
rms.append(listener)
tf.addKeyListener(listener)
tf.addMouseWheelListener(listener)
# Listen to changes in the ROI of imp
rfl = RoiFieldListener(imp, textfields)
Roi.addRoiListener(rfl)
# ... and enable cleanup
ImagePlus.addImageListener(FieldDisabler(rfl, rms))
# Functions for cropping images
cropped = None
cropped_imp = None
def storeRoi(minC, maxC):
if os.path.exists(roi_path):
# Load ROI
with open(path, 'r') as csvfile:
reader = csv.reader(csvfile, delimiter=',', quotechar="\"")
reader.next() # header
same = True
for a, b in izip(minC + maxC, map(int, reader.next()[1:] + reader.next()[1:])):
if a != b:
same = False
# Invalidate any CSV files for features and pointmatches: different cropping
for filename in os.listdir(tgtDir):
if filename.endswith("features.csv") or filename.endswith("pointmatches.csv"):
os.remove(os.path.join(tgtDir, filename))
break
if same:
return
# Store the ROI as crop-roi.csv
with open(roi_path, 'w') as csvfile:
w = csv.writer(csvfile, delimiter=',', quotechar="\"", quoting=csv.QUOTE_NONNUMERIC)
w.writerow(["coords", "x", "y", "z"])
w.writerow(["min"] + map(int, minC))
w.writerow(["max"] + map(int, maxC))
def crop(event):
global cropped, cropped_imp
coords = [int(float(tf.getText())) for tf in textfields]
minC = [max(0, c) for c in coords[0:3]]
maxC = [min(d -1, c) for d, c in izip(Intervals.dimensionsAsLongArray(images[0]), coords[3:6])]
storeRoi(minC, maxC)
print "ROI min and max coordinates"
print minC
print maxC
cropped = [Views.zeroMin(Views.interval(img, minC, maxC)) for img in images]
cropped_imp = showAsStack(cropped, title="cropped")
cropped_imp.setDisplayRange(imp.getDisplayRangeMin(), imp.getDisplayRangeMax())
if cropContinuationFn:
cropContinuationFn(images, minC, maxC, cropped, cropped_imp)
# Buttons to create a ROI and to crop to ROI,
# which when activated enables the fine registration buttons
crop_button = JButton("Crop to ROI")
crop_button.addActionListener(crop)
gc.gridx = 0
gc.gridy += 1
gc.gridwidth = 4
gc.anchor = GBC.WEST
buttons_panel = JPanel()
buttons_panel.add(crop_button)
gb.setConstraints(buttons_panel, gc)
panel.add(buttons_panel)
if independent:
frame = JFrame("Crop by ROI")
frame.getContentPane().add(panel)
frame.pack()
frame.setDefaultCloseOperation(JFrame.DO_NOTHING_ON_CLOSE)
frame.addWindowListener(CloseControl(destroyables=rms + [rfl]))
frame.setVisible(True)
else:
# Re-pack the JFrame
parent = panel.getParent()
while not isinstance(parent, JFrame) and parent is not None:
parent = parent.getParent()
if parent:
frame = parent
frame.pack()
found = False
for wl in frame.getWindowListeners():
if isinstance(wl, CloseControl):
wl.addDestroyables(rms + [rfl])
found = True
break
if not found:
frame.addWindowListener(CloseControl(destroyables=rms + [rfl]))
frame.setVisible(True)
return panel
def maxProjectLastDimension(img, strategy="1by1", chunk_size=0):
last_dimension = img.numDimensions() -1
if "1by1" == strategy:
exe = newFixedThreadPool()
try:
n_threads = exe.getCorePoolSize()
imgTs = [ArrayImgs.unsignedShorts(list(Intervals.dimensionsAsLongArray(img))[:-1]) for i in xrange(n_threads)]
def mergeMax(img1, img2, imgT):
return compute(maximum(img1, img2)).into(imgT)
def hyperSlice(index):
return Views.hyperSlice(img, last_dimension, index)
# The first n_threads mergeMax:
futures = [exe.submit(Task(mergeMax, hyperSlice(i*2), hyperSlice(i*2 +1), imgTs[i]))
for i in xrange(n_threads)]
# As soon as one finishes, merge it with the next available hyperSlice
next = n_threads
while len(futures) > 0: # i.e. not empty
imgT = futures.pop(0).get()
if next < img.dimension(last_dimension):
futures.append(exe.submit(Task(mergeMax, imgT, hyperSlice(next), imgT)))
next += 1
else:
# Run out of hyperSlices to merge
if 0 == len(futures):
return imgT # done
# Merge imgT to each other until none remain
futures.append(exe.submit(Task(mergeMax, imgT, futures.pop(0).get(), imgT)))
finally:
exe.shutdownNow()
else:
# By chunks
imglibtype = img.randomAccess().get().getClass()
# The Converter class
reduce_max = makeCompositeToRealConverter(reducer_class=Math,
reducer_method="max",
reducer_method_signature="(DD)D")
if chunk_size > 0:
# map reduce approach
exe = newFixedThreadPool()
try:
def projectMax(img, minC, maxC, reduce_max):
imgA = ArrayImgs.unsignedSorts(Intervals.dimensionsAsLongArray(imgC))
ImgUtil.copy(ImgView.wrap(convert(Views.collapseReal(Views.interval(img, minC, maxC)), reduce_max.newInstance(), imglibtype), img.factory()), imgA)
return imgA
# The min and max coordinates of all dimensions except the last one
minCS = [0 for d in xrange(last_dimension)]
maxCS = [img.dimension(d) -1 for d in xrange(last_dimension)]
# Process every chunk in parallel
futures = [exe.submit(Task(projectMax, img, minCS + [offset], maxCS + [min(offset + chunk_size, img.dimension(last_dimension)) -1]))
for offset in xrange(0, img.dimension(last_dimension), chunk_size)]
return reduce(lambda f1, f2: compute(maximum(f1.get(), f2.get())).into(f1.get(), futures))
finally:
exe.shutdownNow()
else:
# One chunk: all at once
# Each sample of img3DV is a virtual vector over all time frames at that 3D coordinate
# Reduce each vector to a single scalar, using a Converter
img3DC = convert(Views.collapseReal(img), reduce_max.newInstance(), imglibtype)
imgA = ArrayImgs.unsignedShorts([img.dimension(d) for d in xrange(last_dimension)])
ImgUtil.copy(ImgView.wrap(imgV, img.factory()), imgA)
return imgA
# first take a look at the size and type of each dimension
for d in range(data.numDimensions()):
print "axis d: type: " + str(data.axis(d).type()) + " length: " + str(
data.dimension(d))
img = data.getImgPlus()
xLen = data.dimension(data.dimensionIndex(Axes.X))
yLen = data.dimension(data.dimensionIndex(Axes.Y))
zLen = data.dimension(data.dimensionIndex(Axes.Z))
cLen = data.dimension(data.dimensionIndex(Axes.CHANNEL))
# crop a channel
c0 = ops.transform().crop(
img, Intervals.createMinMax(0, 0, 0, 0, xLen - 1, yLen - 1, 0, zLen - 1))
c0.setName("c0")
# crop both channels at z=12
z12 = ops.transform().crop(
img, Intervals.createMinMax(0, 0, 0, 12, xLen - 1, yLen - 1, cLen - 1, 12))
z12.setName("z12")
# crop channel 0 at z=12
c0z12 = ops.transform().crop(
img, Intervals.createMinMax(0, 0, 0, 12, xLen - 1, yLen - 1, 0, 12))
c0z12.setName("c0z12")
# crop an roi at channel 0, z=12
roiC0z12 = ops.transform().crop(
img, Intervals.createMinMax(150, 150, 0, 12, 200, 200, 0, 12))
from net.imglib2.util import Intervals
from net.imagej.axis import Axes
# first take a look at the size and type of each dimension
for d in range(data.numDimensions()):
print "axis d: type: "+str(data.axis(d).type())+" length: "+str(data.dimension(d))
img=data.getImgPlus()
xLen = data.dimension(data.dimensionIndex(Axes.X))
yLen = data.dimension(data.dimensionIndex(Axes.Y))
zLen = data.dimension(data.dimensionIndex(Axes.Z))
cLen = data.dimension(data.dimensionIndex(Axes.CHANNEL))
# crop a channel
c0=ops.transform().crop(img, Intervals.createMinMax(0, 0, 0,0,xLen-1, yLen-1, 0, zLen-1))
c0.setName("c0")
# crop both channels at z=12
z12=ops.transform().crop(img, Intervals.createMinMax(0,0,0,12, xLen-1, yLen-1, cLen-1, 12))
z12.setName("z12")
# crop channel 0 at z=12
c0z12=ops.transform().crop(img, Intervals.createMinMax(0,0,0,12, xLen-1, yLen-1, 0, 12))
c0z12.setName("c0z12")
# crop an roi at channel 0, z=12
roiC0z12=ops.transform().crop(img, Intervals.createMinMax(150,150,0,12, 200, 200, 0, 12))
roiC0z12.setName("roiC0z12")
#@ OpService ops
#@ SCIFIO scifio
#@ DatasetService datasetService
#@ String input_path
#@ String output_path
#@ UIService ui
import os
from net.imglib2.util import Intervals
from net.imglib2.view import IterableRandomAccessibleInterval
try:
os.unlink(output_path)
except OSError:
pass
input = scifio.datasetIO().open(input_path)
dims = Intervals.dimensionsAsLongArray(input)
output_dims = dims[:-1]
output = ops.create().img(output_dims)
ops.transform().project(IterableRandomAccessibleInterval(output), input,
ops.op('stats.max', input.getImgPlus()),
len(output_dims))
scifio.datasetIO().save(datasetService.create(output), output_path)
ui.show(output)
print("OK")
def export8bitN5(
filepaths,
loadFn,
img_dimensions,
matrices,
name,
exportDir,
interval,
gzip_compression=6,
invert=True,
CLAHE_params=[400, 256, 3.0],
n5_threads=0, # 0 means as many as CPU cores
block_size=[128, 128, 128]):
"""
Export into an N5 volume, in parallel, in 8-bit.
filepaths: the ordered list of filepaths, one per serial section.
loadFn: a function to load a filepath into an ImagePlus.
name: name to assign to the N5 volume.
matrices: the list of transformation matrices (each one is an array), one per section
exportDir: the directory into which to save the N5 volume.
interval: for cropping.
gzip_compression: defaults to 6 as suggested by Saalfeld. 0 means no compression.
invert: Defaults to True (necessary for FIBSEM). Whether to invert the images upon loading.
CLAHE_params: defaults to [400, 256, 3.0]. If not None, the a list of the 3 parameters needed for a CLAHE filter to apply to each image.
n5_threads: defaults to 0, meaning as many as CPU cores.
block_size: defaults to 128x128x128 px. A list of 3 integer numbers, the dimensions of each individual block.
"""
dims = Intervals.dimensionsAsLongArray(interval)
voldims = [dims[0], dims[1], len(filepaths)]
cell_dimensions = [dims[0], dims[1], 1]
def asNormalizedUnsignedByteArrayImg(interval, invert, blockRadius, n_bins,
slope, matrices, index, imp):
sp = imp.getProcessor() # ShortProcessor
# Crop to interval if needed
x = interval.min(0)
y = interval.min(1)
width = interval.max(0) - interval.min(0) + 1
height = interval.max(1) - interval.min(1) + 1
if 0 != x or 0 != y or sp.getWidth() != width or sp.getHeight(
) != height:
sp.setRoi(x, y, width, height)
sp = sp.crop()
if invert:
sp.invert()
CLAHE.run(
ImagePlus("", sp), blockRadius, n_bins, slope, None
) # far less memory requirements than NormalizeLocalContrast, and faster.
minimum, maximum = autoAdjust(sp)
# Transform and convert image to 8-bit, mapping to display range
img = ArrayImgs.unsignedShorts(
sp.getPixels(), [sp.getWidth(), sp.getHeight()])
sp = None
imp = None
# Must use linear interpolation for subpixel precision
affine = AffineTransform2D()
affine.set(matrices[index])
imgI = Views.interpolate(Views.extendZero(img),
NLinearInterpolatorFactory())
imgA = RealViews.transform(imgI, affine)
imgT = Views.zeroMin(Views.interval(imgA, img))
# Convert to 8-bit
imgMinMax = convert2(imgT,
RealUnsignedByteConverter(minimum, maximum),
UnsignedByteType,
randomAccessible=False) # use IterableInterval
aimg = ArrayImgs.unsignedBytes(Intervals.dimensionsAsLongArray(img))
# ImgUtil copies multi-threaded, which is not appropriate here as there are many other images being copied too
#ImgUtil.copy(ImgView.wrap(imgMinMax, aimg.factory()), aimg)
# Single-threaded copy
copier = createBiConsumerTypeSet(UnsignedByteType)
LoopBuilder.setImages(imgMinMax, aimg).forEachPixel(copier)
img = imgI = imgA = imgMinMax = imgT = None
return aimg
blockRadius, n_bins, slope = CLAHE_params
# A CacheLoader that interprets the list of filepaths as a 3D volume: a stack of 2D slices
loader = SectionCellLoader(
filepaths,
asArrayImg=partial(asNormalizedUnsignedByteArrayImg, interval, invert,
blockRadius, n_bins, slope, matrices),
loadFn=loadFn)
# How to preload block_size[2] files at a time? Or at least as many as numCPUs()?
# One possibility is to query the SoftRefLoaderCache.map for its entries, using a ScheduledExecutorService,
# and preload sections ahead for the whole blockSize[2] dimension.
cachedCellImg = lazyCachedCellImg(loader, voldims, cell_dimensions,
UnsignedByteType, BYTE)
exe_preloader = newFixedThreadPool(n_threads=min(
block_size[2], n5_threads if n5_threads > 0 else numCPUs()),
name="preloader")
def preload(cachedCellImg, loader, block_size, filepaths, exe):
"""
Find which is the last cell index in the cache, identify to which block
(given the blockSize[2] AKA Z dimension) that index belongs to,
and concurrently load all cells (sections) that the Z dimension of the blockSize will need.
If they are already loaded, these operations are insignificant.
"""
try:
# The SoftRefLoaderCache.map is a ConcurrentHashMap with Long keys, aka numbers
cache = cachedCellImg.getCache()
f1 = cache.getClass().getDeclaredField(
"cache") # LoaderCacheAsCacheAdapter.cache
f1.setAccessible(True)
softCache = f1.get(cache)
cache = None
f2 = softCache.getClass().getDeclaredField(
"map") # SoftRefLoaderCache.map
f2.setAccessible(True)
keys = sorted(f2.get(softCache).keySet())
if 0 == len(keys):
return
first = max(0, keys[-1] - (keys[-1] % block_size[2]))
last = min(len(filepaths), first + block_size[2]) - 1
keys = None
syncPrintQ("### Preloading %i-%i ###" % (first, last))
futures = []
for index in xrange(first, last + 1):
futures.append(
exe.submit(TimeItTask(softCache.get, index, loader)))
softCache = None
# Wait for all
loaded_any = False
count = 0
while len(futures) > 0:
r, t = futures.pop(0).get() # waits for the image to load
if t > 1000: # in miliseconds. Less than this is for sure a cache hit, more a cache miss and reload
loaded_any = True
r = None
# t in miliseconds
syncPrintQ("preloaded index %i in %f ms" % (first + count, t))
count += 1
if not loaded_any:
syncPrintQ("Completed preloading %i-%i" %
(first, first + block_size[2] - 1))
except:
syncPrintQ(sys.exc_info())
preloader = Executors.newSingleThreadScheduledExecutor()
preloader.scheduleWithFixedDelay(
RunTask(preload, cachedCellImg, loader, block_size, filepaths,
exe_preloader), 10, 60, TimeUnit.SECONDS)
try:
syncPrint("N5 directory: " + exportDir + "\nN5 dataset name: " + name +
"\nN5 blockSize: " + str(block_size))
writeN5(cachedCellImg,
exportDir,
name,
block_size,
gzip_compression_level=gzip_compression,
n_threads=n5_threads)
finally:
preloader.shutdown()
exe_preloader.shutdown()
# @Dataset data
# @Dataset mask
# @OUTPUT Dataset output
# Given a mask (binary image) and a raw image, remove background pixel from raw by
# keeping only those in the mask (different from 0).
# Note : As specified by @stelfrich on Gitter, the particular case when foreground pixel
# are 1 and background pixels are 0 can be simpler to write with a multiplication of the two
# images.
from net.imglib2.util import Intervals
# Check dimensions are the same for 'data' and 'mask'
if not Intervals.equalDimensions(data, mask):
raise Exception("Dimensions from input dataset does not match.")
# Create the cursors
output = data.duplicate()
targetCursor = output.localizingCursor()
dataRA = data.randomAccess()
maskRA = mask.randomAccess()
# Iterate over each pixels of the datasets
while targetCursor.hasNext():
targetCursor.fwd()
dataRA.setPosition(targetCursor)
maskRA.setPosition(targetCursor)
if maskRA.get().get() == 0:
def get(self, index):
img = self.asArrayImg(index, self.loadFn(self.filepaths[index]))
dims = Intervals.dimensionsAsLongArray(img)
return Cell(
list(dims) + [1], [0] * img.numDimensions() + [index],
img.update(None))
# by using the 'transformed' float array.
# We compute the bounds by, for every corner, checking if the floor of each dimension
# of a corner coordinate is smaller than the previously found minimum value,
# and by checking if the ceil of each corner coordinate is larger than the
# previously found value, packing the new pair of minimum and maximum values
# into the list of pairs that is 'bounds'.
# Notice the min coordinates can have negative values, as the rotated image
# has pixels now somewhere to the left and up from the top-left 0,0,0 origin
# of coordinates. That's why we use Views.zeroMin, to ensure that downstream
# uses of the transformed image see it as fitting within bounds that start at 0,0,0.
bounds = repeat(
(sys.maxint, 0)
) # initial upper- and lower-bound values for min, max to compare against
transformed = zeros(img.numDimensions(), 'f')
for corner in product(*zip(repeat(0), Intervals.maxAsLongArray(img))):
rotation.apply(corner, transformed)
bounds = [(min(vmin, int(floor(v))), max(vmax, int(ceil(v))))
for (vmin, vmax), v in zip(bounds, transformed)]
minC, maxC = map(list, zip(*bounds)) # transpose list of lists
imgRot2dFit = IL.wrap(Views.zeroMin(Views.interval(rotated, minC, maxC)),
imp.getTitle() + " - rot2dFit")
imgRot2dFit.show()
matrix = rotation.getRowPackedCopy()
pprint([list(matrix[i:i + 4]) for i in xrange(0, 12, 4)])
converted = ops.convert().float32(data)
# Get the first frame (TODO: find a more convenient way !)
t_dim = data.dimensionIndex(Axes.TIME)
interval_start = []
interval_end = []
for d in range(0, data.numDimensions()):
if d != t_dim:
interval_start.append(0)
interval_end.append(data.dimension(d) - 1)
else:
interval_start.append(0)
interval_end.append(0)
intervals = interval_start + interval_end
intervals = Intervals.createMinMax(*intervals)
first_frame = ops.transform().crop(converted, intervals)
# Allocate output memory (wait for hybrid CF version of slice)
subtracted = ops.create().img(converted)
# Create the op
sub_op = ops.op("math.subtract", first_frame, first_frame)
# Setup the fixed axis
fixed_axis = [d for d in range(0, data.numDimensions()) if d != t_dim]
# Run the op
ops.slice(subtracted, converted, sub_op, fixed_axis)
