python-gamera 3.3.2-2 / usr / share / pyshared / gamera / plugins / segmentation.py

#
# Copyright (C) 2001-2005 Ichiro Fujinaga, Michael Droettboom, and Karl MacMillan
#
# This program is free software; you can redistribute it and/or
# modify it under the terms of the GNU General Public License
# as published by the Free Software Foundation; either version 2
# of the License, or (at your option) any later version.
#
# This program is distributed in the hope that it will be useful,
# but WITHOUT ANY WARRANTY; without even the implied warranty of
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
# GNU General Public License for more details.
# 
# You should have received a copy of the GNU General Public License
# along with this program; if not, write to the Free Software
# Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA.
#

from gamera.plugin import *
from gamera import util
import _segmentation


class Segmenter(PluginFunction):
    self_type = ImageType([ONEBIT])
    return_type = ImageList("ccs")
    doc_examples = [(ONEBIT,)]

class cc_analysis(Segmenter):
    """
    Performs connected component analysis on the image.

    This algorithm assumes 8-connected components, meaning any two
    pixels are considered "connected" if they are adjacent in any
    direction, including diagonally.

    The original image will have all of its pixels "labeled" with a
    number representing each connected component.  This is so the
    connected components can share data with their source image and
    makes things much more efficient.

    Returns a list of ccs found in the image.  Since all the CC's
    share the same data with the original image, changing the CC's
    will affect the original.  If you do not want this behavior, use
    the image_copy_ function on each of the CCs::

      ccs = [x.image_copy() for x in ccs]

    .. _image_copy: utility.html#image-copy
    """
    pass


class cc_and_cluster(Segmenter):
    """
    Performs connected component analysis using cc_analysis_ and then
    clusters the CC's according to their similarity.

    TODO: We need some more detailed documentation here.
    """
    pure_python = True
    args = Args([Float('ratio', default = 1.0), Int('distance', default=2)])
    return_type = ImageList("ccs")
    def __call__(image, ratio = 1.0, distance = 2):
        from gamera import cluster
        cc = image.cc_analysis()
        return cluster.cluster(cc, ratio, distance)
    __call__ = staticmethod(__call__)
    doc_examples = [(ONEBIT,)]

class splitx(Segmenter):
    """
    Splits an image vertically.

    The split point is determined automatically by finding a valley in
    the projections near *center*.

    This function is overloaded to work both with a single value
    and a list of splitting point candidates as input.
    """
    args = Args([FloatVector("center", default=[0.5])])
    doc_examples = [(ONEBIT,)]
    def __call__(self, center=0.5):
       if not util.is_sequence(center):
          return _segmentation.splitx(self, [center])
       else:
          return _segmentation.splitx(self, center)
    __call__ = staticmethod(__call__)
    author = "Michael Droettboom, Karl MacMillan and Christoph Dalitz"

class splitx_max(Segmenter):
    """Splits an image vertically.

    The split point is determined automatically by finding a peak in
    the projections near *center*.

    This function is overloaded to work both with a single value and a
    list of splitting point canidates as input.
    """
    args = Args([FloatVector("center", default=[0.5])])
    def __call__(self, center=0.5):
       if not util.is_sequence(center):
          return _segmentation.splitx_max(self, [center])
       else:
          return _segmentation.splitx_max(self, center)
    __call__ = staticmethod(__call__)
    author = "Michael Droettboom, Karl MacMillan and Christoph Dalitz"

class splity(Segmenter):
    """
    Splits an image horizontally.

    The split point is determined automatically by finding a valley in
    the projections near *center*.

    This function is overloaded to work both with a single value and a
    list of splitting point canidates as input.
    """
    args = Args([FloatVector("center", default=[0.5])])
    def __call__(self, center=[0.5]):
       if not util.is_sequence(center):
          return _segmentation.splity(self, [center])
       else:
          return _segmentation.splity(self, center)
    __call__ = staticmethod(__call__)
    author = "Michael Droettboom, Karl MacMillan and Christoph Dalitz"

class splitx_base(Segmenter):
    pure_python = True
    return_type = ImageList("splits")
    
class splitx_left(splitx_base):
    """
    Splits an image vertically.

    The split point is determined automatically by finding a valley in
    the projections near the left of the image.
    """
    _center = 0.25
    def __call__(self):
        return self.splitx(0.25)
    __call__ = staticmethod(__call__)

class splitx_right(splitx_base):
    """
    Splits an image vertically.

    The split point is determined automatically by finding a valley in
    the projections near the right of the image.
    """
    _center = 0.75
    def __call__(self):
        return self.splitx(0.75)
    __call__ = staticmethod(__call__)

class splity_base(Segmenter):
    pure_python = True
    return_type = ImageList("splits")
    
class splity_top(splity_base):
    """
    Splits an image horizontally.

    The split point is determined automatically by finding a valley in
    the projections near the top of the image.
    """
    _center = 0.25
    def __call__(self):
        return self.splity(0.25)
    __call__ = staticmethod(__call__)

class splity_bottom(splity_base):
    """
    Splits an image horizontally.

    The split point is determined automatically by finding a valley in
    the projections near the bottom of the image.
    """
    _center = 0.75
    def __call__(self):
        return self.splity(0.75)
    __call__ = staticmethod(__call__)

# connected-component filters

def filter_wide(ccs, max_width):
    tmp = []
    for x in ccs:
        if x.ncols > max_width:
            x.fill_white()
        else:
            tmp.append(x)
    return tmp

def filter_narrow(ccs, min_width):
    tmp = []
    for x in ccs:
        if x.ncols < min_width:
            x.fill_white()
        else:
            tmp.append(x)
    return tmp

def filter_tall(ccs, max_height):
    tmp = []
    for x in ccs:
        if x.nrows > max_height:
            x.fill_white()
        else:
            tmp.append(x)
    return tmp

def filter_short(ccs, min_height):
    tmp = []
    for x in ccs:
        if x.nrows < min_height:
            x.fill_white()
        else:
            tmp.append(x)
    return tmp

def filter_small(ccs, min_size):
    tmp = []
    for x in ccs:
        if x.nrows < min_size or x.ncols < min_size:
            x.fill_white()
        else:
            tmp.append(x)
    return tmp

def filter_large(ccs, max_size):
    tmp = []
    for x in ccs:
        if x.nrows > max_size or x.ncols > max_size:
            x.fill_white()
        else:
            tmp.append(x)
    return tmp

def filter_black_area_small(ccs, min_size):
    tmp = []
    for x in ccs:
        if x.black_area()[0] < min_size:
            x.fill_white()
        else:
            tmp.append(x)
    return tmp

def filter_black_area_large(ccs, max_size):
    tmp = []
    for x in ccs:
        if x.black_area()[0] > max_size:
            x.fill_white()
        else:
            tmp.append(x)
    return tmp

class SegmentationModule(PluginModule):
    category = "Segmentation"
    cpp_headers=["segmentation.hpp"]
    functions = [cc_analysis, cc_and_cluster, splitx, splity,
                 splitx_left, splitx_right, splity_top, splity_bottom,
                 splitx_max]
    author = "Michael Droettboom and Karl MacMillan"
    url = "http://gamera.sourceforge.net/"

module = SegmentationModule()

del Segmenter
del splitx_base
del splity_base