Source code for diamondback.filters.RankFilter

""" **Description**

        A rank filter realizes a nonlinear morphological operator consuming an
        incident signal, sorting, indexing, and selecting over a sliding window
        of a specified order, and producing a reference signal.

        Simple morphological operations include dilation, which defines a rank
        index equal to zero, erosion, which defines a rank index equal to order,
        and median, which defines a rank index equal to order divided by two
        for even order.  Compound morphological operations include close,
        consisting of sequential dilation and erosion, and open, consisting of
        sequential erosion and dilation.  An order and rank are specified.

        .. math::

            y_{n} = sort(\ x_{n-N+1\ :\ n}\ )[\ i\ ]

    **Example**

        ::

            from diamondback import RankFilter
            import numpy


            # Create an instance with rank and order.

            obj = RankFilter( rank = 4, order = 4 )

            # Filter an incident signal.

            x = numpy.concatenate( ( numpy.ones( 1 ), numpy.zeros( 10 ), numpy.ones( 4 ), numpy.zeros( 2 ), numpy.ones( 5 ), numpy.zeros( 6 ) ) )

            obj.reset( x[ 0 ] )

            y = obj.filter( x )

    **License**

        `BSD-3C.  <https://github.com/larryturner/diamondback/blob/master/license>`_

        © 2018 - 2021 Larry Turner, Schneider Electric Industries SAS. All rights reserved.

    **Author**

        Larry Turner, Schneider Electric, Analytics & AI, 2018-01-31.

    **Definition**

"""

from diamondback.filters.FirFilter import FirFilter
import numpy
import typing


[docs]class RankFilter( FirFilter ) : """ Rank filter. """ @property def rank( self ) : """ rank : int - in [ 0, order ]. """ return self._rank @rank.setter def rank( self, rank : int ) : if ( ( rank < 0 ) or ( rank > ( len( self.s ) - 1 ) ) ) : raise ValueError( f'Rank = {rank}' ) self._rank = rank
[docs] def __eq__( self, other : typing.Any ) -> bool : """ Equal. Arguments : other : typing.Any. Returns : equal : bool. """ return ( ( super( ).__eq__( other ) ) and ( self.rank == other.rank ) )
def __init__( self, rank : int, order : int ) -> None : """ Initialize. Arguments : rank : int - in [ 0, order ]. order : int. """ if ( ( rank < 0 ) or ( rank > order ) ) : raise ValueError( f'Rank = {rank} Order = {order}' ) super( ).__init__( numpy.ones( order + 1 ) / ( order + 1 ) ) self._rank = rank
[docs] def filter( self, x : typing.Union[ typing.List, numpy.ndarray ] ) -> numpy.ndarray : """ Filters an incident signal and produces a reference signal. Arguments : x : typing.Union[ typing.List, numpy.ndarray ] - incident signal. Returns : y : numpy.ndarray - reference signal. """ if ( ( not numpy.isscalar( x ) ) and ( not isinstance( x, numpy.ndarray ) ) ) : x = numpy.array( list( x ) ) if ( ( len( x.shape ) != 1 ) or ( len( x ) == 0 ) ) : raise ValueError( f'X = {x}' ) y = numpy.zeros( len( x ), type( x[ 0 ] ) ) for ii in range( 0, len( x ) ) : self.s[ 0 ] = x[ ii ] y[ ii ] = self.s[ numpy.argsort( abs( self.s ) ) ][ self.rank ] if ( len( self.s ) > 1 ) : self.s[ 1 : ] = self.s[ : -1 ] return y