Constructor Detail

• StructuringElements

  public StructuringElements()

Method Detail

• disk

  public static final List<Shape> disk(long radius,
                                       int dimensionality)

  Generates a centered disk flat structuring element for morphological operations.

  The structuring element (strel) is returned as a List of Shapes, for structuring elements can be decomposed to yield a better performance. In 2D, the disk strel can be approximated by several periodic lines. The resulting strel is only an approximation of a disk, and this method offers a parameter to select the level of approximation. For other dimensionalities, no optimization are available yet and the parameter is ignored.

  This methods relies on heuristics to determine automatically what decomposition level to use.

  Parameters:

  radius - the radius of the disk, so that it extends over 2 × radius + 1 in all dimensions

  dimensionality - the dimensionality of the target problem.

  Returns:

  a disk structuring element as a new list of Shapes.

• disk

  public static final List<Shape> disk(long radius,
                                       int dimensionality,
                                       int decomposition)

  Generates a centered disk flat structuring element for morphological operations.

  The structuring element (strel) is returned as a List of Shapes, for structuring elements can be decomposed to yield a better performance. In 2D, the disk strel can be approximated by several periodic lines. The resulting strel is only an approximation of a disk, and this method offers a parameter to select the level of approximation. For other dimensionalities, no optimization are available yet and the parameter is ignored.

  Parameters:

  radius - the radius of the disk, so that it extends over 2 × radius + 1 in all dimensions

  dimensionality - the dimensionality of the target problem.

  decomposition - the decomposition to use. Only values 0, 4, 6 and 8 are accepted:

  0. No approximation is made and a full dimension-generic disk is returned.
  4. The disk is decomposed in 4 periodic lines, plus in some cases 2 horizontal lines.
  6. The disk is decomposed in 6 periodic lines, plus in some cases 2 horizontal lines.
  8. The disk is decomposed in 8 periodic lines, plus in some cases 2 horizontal lines.

  This parameter is ignored for dimensionality other than 2.

  Returns:

  a disk structuring element as a new list of Shapes.

• square

  public static final List<Shape> square(int radius,
                                         int dimensionality,
                                         boolean decompose)

  Generates a centered square flat structuring element for morphological operations.

  This method specify the square size using its radius to comply to sibling methods. The extend of the generated square is 2 × radius + 1 in all dimensions.

  The structuring element (strel) is returned as a List of Shapes, for Structuring elements can be decomposed to yield a better performance. The square strel can be decomposed in a succession of orthogonal lines and yield the exact same results on any of the morphological operations. Because the decomposition becomes dimension-specific, the dimensionality of the target problem must be specified. Warning: Undesired effects will occur if the specified dimensionality and target dimensionality do not match. Non-decomposed version are dimension-generic.

  Parameters:

  radius - the radius of the square.

  dimensionality - the dimensionality of the target problem.

  decompose - if true, the structuring element will be optimized through decomposition.

  Returns:

  a new structuring element, as a list of Shapes.

• square

  public static final List<Shape> square(int radius,
                                         int dimensionality)

  Generates a centered square flat structuring element for morphological operations.

  This method specify the square size using its radius to comply to sibling methods. The extend of the generated square is 2 × radius + 1 in all dimensions.

  The structuring element (strel) is returned as a List of Shapes, for Structuring elements can be decomposed to yield a better performance. The square strel can be decomposed in a succession of orthogonal lines and yield the exact same results on any of the morphological operations. Because the decomposition becomes dimension-specific, the dimensionality of the target problem must be specified. Warning: Undesired effects will occur if the specified dimensionality and target dimensionality do not match. Non-decomposed version are dimension-generic.

  This method determines whether it is worth returning a decomposed strel based on simple heuristics.

  Parameters:

  radius - the radius of the square.

  dimensionality - the dimensionality of the target problem.

  Returns:

  a new structuring element, as a list of Shapes.

• rectangle

  public static final List<Shape> rectangle(int[] halfSpans,
                                            boolean decompose)

  Generates a symmetric, centered, rectangular flat structuring element for morphological operations.

  The structuring element (strel) is returned as a List of Shapes, for Structuring elements can be decomposed to yield a better performance. The rectangle strel can be decomposed in a succession of orthogonal lines and yield the exact same results on any of the morphological operations.

  Parameters:

  halfSpans - an int[] array containing the half-span of the symmetric rectangle in each dimension. The total extent of the rectangle will therefore be 2 × halfSpan[d] + 1 in each dimension.

  decompose - if true, the strel will be returned as a List of HorizontalLineShape, indeed performing the rectangle decomposition. If false, the list will be made of a single CenteredRectangleShape.

  Returns:

  the desired structuring element, as a List of Shapes.

• rectangle

  public static final List<Shape> rectangle(int[] halfSpans)

  Generates a symmetric, centered, rectangular flat structuring element for morphological operations.

  The structuring element (strel) is returned as a List of Shapes, for Structuring elements can be decomposed to yield a better performance. The rectangle strel can be decomposed in a succession of orthogonal lines and yield the exact same results on any of the morphological operations. This method uses a simple heuristic to decide whether to decompose the rectangle or not.

  Parameters:

  halfSpans - an int[] array containing the half-span of the symmetric rectangle in each dimension. The total extent of the rectangle will therefore be 2 × halfSpan[d] + 1 in each dimension.

  Returns:

  the desired structuring element, as a List of Shapes.

• diamond

  public static final List<Shape> diamond(int radius,
                                          int dimensionality)

  Generates a centered flat diamond structuring element for morphological operations.

  The structuring element (strel) is returned as a List of Shapes, for Structuring elements can be decomposed to yield a better performance. Because the decomposition is dimension-specific, this methods requires it to be specified. Warning: using a structuring element built with the wrong dimension can and will lead to undesired (and sometimes hard to detect) defects in subsequent morphological operations.

  The diamond strel can be effectively decomposed in 2D (and 1D) using the logarithmic decomposition in extreme sets, as explained in [1]. For other dimensions, the theorem does not hold (even in practice), and we have to fall back on a linear decomposition, still very effective (see [1] as well).

  Parameters:

  radius - the desired radius of the diamond structuring element. The strel will extend over 2 × radius + 1 in all dimensions.

  dimensionality - the target dimensionality this structuring element will be used with. A structuring element build for one dimension will not work properly for any other dimensions.

  Returns:

  the structuring element as a list of Shapes.

  See Also:

  [1] Rein van den Boomgard and Richard van Balen, Methods for Fast Morphological Image Transforms Using Bitmapped Binary Images, CVGIP: Models and Image Processing, vol. 54, no. 3, May 1992, pp. 252-254.

• diamond

  public static final List<Shape> diamond(int radius,
                                          int dimensionality,
                                          boolean decompose)

  Generates a centered flat diamond structuring element for morphological operations.

  The structuring element (strel) is returned as a List of Shapes, for Structuring elements can be decomposed to yield a better performance. Because the decomposition is dimension-specific, this methods requires it to be specified. Warning: using a structuring element built with the wrong dimension can and will lead to undesired (and sometimes hard to detect) defects in subsequent morphological operations. Non-optimized versions of this strel are dimension-generic.

  The diamond strel can be effectively decomposed in 2D (and 1D) using the logarithmic decomposition in extreme sets, as explained in [1]. For other dimensions, the theorem does not hold (even in practice), and we have to fall back on a linear decomposition, still very effective (see [1] as well).

  Parameters:

  radius - the desired radius of the diamond structuring element. The strel will extend over 2 × radius + 1 in all dimensions.

  dimensionality - the target dimensionality this structuring element will be used with. A structuring element build for one dimension will not work properly for any other dimensions.

  decompose - if true, this strel will be optimized through decomposition.

  Returns:

  the structuring element as a list of Shapes.

  See Also:

  [1] Rein van den Boomgard and Richard van Balen, Methods for Fast Morphological Image Transforms Using Bitmapped Binary Images, CVGIP: Models and Image Processing, vol. 54, no. 3, May 1992, pp. 252-254.

• periodicLine

  public static final Shape periodicLine(long span,
                                         int[] increments)

  Creates a new periodic line structuring element, that will iterate over 2 × span + 1 pixels as follow:

   position - span x increments,
   ...
   position - 2 × increments,
   position - increments,
   position,
   position + increments,
   position + 2 × increments,
   ...
   position + span x increments
   

  The importance of periodic lines is explained in [1].

  Parameters:

  span - the span of the neighborhood, so that it will iterate over 2 × span + 1 pixels.

  increments - the values by which each element of the position vector is to be incremented when iterating.

  See Also:

  [1] Jones and Soilles.Periodic lines: Definition, cascades, and application to granulometries. Pattern Recognition Letters (1996) vol. 17 (10) pp. 1057-1063.