def radon(image, stepsize=2, maskrad=None):
from multiprocessing import Queue, Process
t0 = time.time()
### prepare mask
if maskrad is None:
maskrad = image.shape[0] / 2
blackcircle = imagefun.filled_circle(image.shape, maskrad)
mask = 1 - blackcircle
nsteps = int(math.ceil(180 / stepsize))
queuelist = []
for row in range(nsteps):
angle = -row * stepsize
queue = Queue()
queuelist.append(queue)
#below is equivalent to "project(image, row, angle, mask, queue)"
proc = Process(target=project, args=(image, row, angle, mask, queue))
proc.start()
proc.join()
### assemble radon image
radonimage = numpy.zeros((nsteps, image.shape[0]))
for queue in queuelist:
row, line = queue.get()
radonimage[row, :] = line
#radonlr = numpy.fliplr(radonimage)
#radonimage = numpy.vstack((radonimage, radonlr))
imagefile.arrayToJpeg(radonimage, "radonimage.jpg", msg=False)
print "Multi radon completed in %s" % (apDisplay.timeString(time.time() -
t0))
return radonimage
def radon(image, stepsize=2, maskrad=None): from multiprocessing import Queue, Process t0 = time.time() ### prepare mask if maskrad is None: maskrad = image.shape[0]/2 blackcircle = imagefun.filled_circle(image.shape, maskrad) mask = 1 - blackcircle nsteps = int(math.ceil(180/stepsize)) queuelist = [] for row in range(nsteps): angle = -row*stepsize queue = Queue() queuelist.append(queue) #below is equivalent to "project(image, row, angle, mask, queue)" proc = Process(target=project, args=(image, row, angle, mask, queue)) proc.start() proc.join() ### assemble radon image radonimage = numpy.zeros( (nsteps, image.shape[0]) ) for queue in queuelist: row, line = queue.get() radonimage[row, :] = line #radonlr = numpy.fliplr(radonimage) #radonimage = numpy.vstack((radonimage, radonlr)) imagefile.arrayToJpeg(radonimage, "radonimage.jpg", msg=False) print "Multi radon completed in %s"%(apDisplay.timeString(time.time() - t0)) return radonimage
b = ndimage.shift(b, shift=shift, mode='wrap', order=1)
bnoise = b + noiselevel*numpy.random.random(shape)
bnoise = ndimage.median_filter(bnoise, size=2)
bnoise = imagenorm.normStdev(bnoise)
return bnoise
if __name__ == "__main__":
u = mem.used()
stepsize = 1.0
shape = (256,256)
noiselevel = 2.0
a = createDots(angle=0., noiselevel=noiselevel, shape=shape)
b = createDots(angle=-30., noiselevel=noiselevel, shape=shape, shift=(3,4))
imagefile.arrayToJpeg(a, "imagea.jpg")
imagefile.arrayToJpeg(b, "imageb.jpg")
ar = radonTransform(a, stepsize)
br = radonTransform(b, stepsize)
imagefile.arrayToJpeg(ar, "radona.jpg")
imagefile.arrayToJpeg(br, "radonb.jpg")
#print "%.3f"%(a[1,0])
#print "%.3f"%(b[1,0])
for radius in range(0,11,2):
e = radermacher.radonshift(ar,br,radius)
e = e / (ar.std()*br.std()) # result is not normalized properly
val = e.argmax()
y = val%e.shape[1]
x = int(val/e.shape[1])
def findEllipseEdge(self, fftarray, mindef=None, maxdef=None): if mindef is None: mindef = self.params['mindef'] if maxdef is None: maxdef = self.params['maxdef'] minpeaks = ctftools.getCtfExtrema(maxdef, self.freq*1e10, self.ctfvalues['cs'], self.ctfvalues['volts'], 0.0, numzeros=3, zerotype="peaks") minEdgeRadius = minpeaks[0] maxvalleys = ctftools.getCtfExtrema(mindef, self.freq*1e10, self.ctfvalues['cs'], self.ctfvalues['volts'], 0.0, numzeros=3, zerotype="valleys") minEdgeRadius = (maxvalleys[0]+minpeaks[0])/2 maxEdgeRadius = maxvalleys[2] minCircum = math.ceil(2 * minEdgeRadius * math.pi) minEdges = int(minCircum) maxCircum = math.ceil(2 * maxEdgeRadius * math.pi) maxEdges = 2*int(maxCircum) dograd = (minpeaks[1] - minpeaks[0])/3.0 dogarray = apDog.diffOfGauss(fftarray, dograd, k=1.2) #dogarray = fftarray print "Edge range: ", minEdges, maxEdges edges = canny.canny_edges(dogarray, low_thresh=0.01, minEdgeRadius=minEdgeRadius, maxEdgeRadius=maxEdgeRadius, minedges=minEdges, maxedges=maxEdges) #minedges=2500, maxedges=15000, invedges = numpy.fliplr(numpy.flipud(edges)) self.edgeMap = numpy.logical_and(edges, invedges) edges = self.edgeMap * (dogarray.max()/self.edgeMap.max()) imagedata = dogarray.copy() + edges imagefile.arrayToJpeg(imagedata, "edges.jpg") edgeThresh = 3 ellipseParams = ransac.ellipseRANSAC(self.edgeMap, edgeThresh, maxRatio=self.maxRatio) if ellipseParams is None: return None fitEllipse1 = ransac.generateEllipseRangeMap2(ellipseParams, edgeThresh, self.edgeMap.shape) fitEllipse2 = ransac.generateEllipseRangeMap2(ellipseParams, edgeThresh*3, self.edgeMap.shape) outlineEllipse = fitEllipse2 - fitEllipse1 # image is draw upside down #outlineEllipse = numpy.flipud(outlineEllipse) imagedata = imagedata + 0.5*outlineEllipse*imagedata.max() if self.debug is True: from matplotlib import pyplot pyplot.clf() pyplot.imshow(imagedata) pyplot.gray() pyplot.show() self.ellipseParams = ellipseParams #self.convertEllipseToCtf(node=3) return ellipseParams
b = ndimage.shift(b, shift=shift, mode='wrap', order=1) bnoise = b + noiselevel*numpy.random.random(shape) bnoise = ndimage.median_filter(bnoise, size=2) bnoise = imagenorm.normStdev(bnoise) return bnoise if __name__ == "__main__": u = mem.used() stepsize = 1.0 shape = (256,256) noiselevel = 2.0 a = createDots(angle=0., noiselevel=noiselevel, shape=shape) b = createDots(angle=-30., noiselevel=noiselevel, shape=shape, shift=(3,4)) imagefile.arrayToJpeg(a, "imagea.jpg") imagefile.arrayToJpeg(b, "imageb.jpg") ar = radonTransform(a, stepsize) br = radonTransform(b, stepsize) imagefile.arrayToJpeg(ar, "radona.jpg") imagefile.arrayToJpeg(br, "radonb.jpg") #print "%.3f"%(a[1,0]) #print "%.3f"%(b[1,0]) for radius in range(0,11,2): e = radermacher.radonshift(ar,br,radius) e = e / (ar.std()*br.std()) # result is not normalized properly val = e.argmax() y = val%e.shape[1] x = int(val/e.shape[1])