faster version of smoother using the scipy gaussian filter to get the diffused microstructure and vectorized sharpening of the diffused microstructure

This commit is contained in:
Pratheek Shanthraj 2013-06-21 16:59:49 +00:00
parent 45951e27c5
commit 2e8756b724
1 changed files with 9 additions and 32 deletions

View File

@ -3,6 +3,7 @@
import os,sys,string,re,math,numpy import os,sys,string,re,math,numpy
from optparse import OptionParser, OptionGroup, Option, SUPPRESS_HELP from optparse import OptionParser, OptionGroup, Option, SUPPRESS_HELP
from scipy import ndimage
#-------------------------------------------------------------------------------------------------- #--------------------------------------------------------------------------------------------------
class extendedOption(Option): class extendedOption(Option):
@ -160,38 +161,14 @@ for file in files:
i/info['grid'][0] /info['grid'][1]] = item i/info['grid'][0] /info['grid'][1]] = item
i += 1 i += 1
for i in range(options.N): maxMicro = numpy.zeros(info['grid'])
active = [] microstructureNew = numpy.zeros(info['grid'],'i')
for z in xrange(info['grid'][2]): for i in range(numpy.amax(microstructure)):
for y in xrange(info['grid'][1]): diffusedMicro = ndimage.filters.gaussian_filter((microstructure == i).astype(float),numpy.sqrt(options.N))
for x in xrange(info['grid'][0]): microstructureNew = numpy.where(diffusedMicro > maxMicro,i,microstructureNew)
me = microstructure[x,y,z] maxMicro = numpy.where(diffusedMicro > maxMicro,diffusedMicro,maxMicro)
others = me*numpy.ones(1+len(neighborhood),'i')
hot = False microstructure = microstructureNew
o = 0
for offset in neighborhood:
otherX = (x+offset[0])%info['grid'][0]
otherY = (y+offset[1])%info['grid'][1]
otherZ = (z+offset[2])%info['grid'][2]
other = microstructure[otherX,otherY,otherZ]
o += 1
others[o] = other
if other != me:
hot = True
if hot:
active.insert(0,numpy.array([x,y,z,0,0],'i')) # remember current position, best candidate, and best change
# append might be not a good idea in linked lists. try to put at the start, not end!
for o in xrange(len(others)):
howMany = numpy.array(others[1:] == others[o],'i').sum() # count number of particular others in neighborhood
if active[0][3] < howMany:
active[0][3:5] = [howMany,others[o]]
for spot in active:
microstructure[spot[0],spot[1],spot[2]] = spot[4]
formatwidth = int(math.floor(math.log10(microstructure.max())+1)) formatwidth = int(math.floor(math.log10(microstructure.max())+1))