Attribute Catalog

This catalog lists the scalar attributes exposed through mtlearn.morphology.AttributeType and the groups exposed through mtlearn.morphology.AttributeGroup: 133 scalar attributes and nine groups. Names, groups, and definitions follow the mmcfilters v5.2.0 attribute catalog.

Attributes are computed per morphology-tree node slot. They are not image arrays. The usual result layout is:

attribute_index, values = morphology.compute_attributes(tree, attributes)
column = attribute_index["AREA"]
area_by_node_slot = values[:, column]

Attribute buffers default to np.float32. Pass dtype=np.float64 to compute_attributes or compute_single_attribute when double-precision attribute values are needed. Deterministic attribute filters accept both float32 and float64 attribute vectors.

Use tree.num_internal_node_slots when allocating criteria, scores, or any node-indexed array that will be passed back to filtering code.

Terminology

A morphology tree represents connected components or shapes by tree nodes. The proper part of a node is the set of image pixels assigned directly to that node. Each pixel belongs to one proper part and is identified by its row-major linear index. Attribute vectors are indexed by tree nodes.

A node slot is the dense integer row used internally for one internal tree node. Attribute arrays keep one row per slot so the same id can be reused by filters, scores, and reconstruction code. In ordinary trees these slots correspond to live internal nodes; after advanced topology edits, some slots may be kept for stable indexing.

The support of a node is the set of image pixels represented by that node: the pixels directly owned by the node plus the pixels represented by its descendant nodes. In a max-tree or min-tree, this support is the connected component associated with that tree node at its altitude.

Attribute Groups

Groups are convenience requests. They expand to canonical scalar attributes before computation or CFP parameter construction. Each scalar is listed once below, with its group memberships. ALL includes every listed scalar.

Group	Meaning
GRAY_LEVEL	Node valuation, gray-level mass, contrast, height, mean, and variance.
SHAPE	Size, bounding box, moment, bitquad, contour, and selected shape descriptors.
MOMENTS	Central moments, Hu invariants, and moment-derived descriptors.
BOUNDARY	Bitquad and contour descriptors.
TREE_TOPOLOGY	Parent/child hierarchy descriptors.
DIST_TRANSF	Descriptors of the approximate contour-distance field.
DIST_TRANSF_EXACT	Descriptors of the exact Euclidean contour-distance field.
FILLED_SHAPE	Geometry of the region enclosed by the external boundary, including holes.
ALL	Every public scalar attribute.

shape_attributes = morphology.expand_attribute_group(
    morphology.AttributeGroup.SHAPE,
)

Contracts

The catalog uses three contracts:

Contract	Meaning
Altitude-aware	Reads node altitude or gray-level values from a valued tree.
Topology/support	Uses node support, pixel coordinates, contours, or adjacency information.
Tree topology	Reads only parent/child relations in the hierarchy.

Gray-Level Attributes

Attribute	Groups	Contract	Use
GRAY_LEVEL	GRAY_LEVEL	Altitude-aware	Node valuation: the gray level associated with the node.
VOLUME	GRAY_LEVEL	Altitude-aware	Sum of altitude-weighted support contributions over the node subtree.
RELATIVE_VOLUME	GRAY_LEVEL	Altitude-aware	Recursive contrast volume R(n) = area(n) + sum_c [R(c) + area(c) * abs(altitude(c) - altitude(n))] over the direct children c.
GRAY_LEVEL_HEIGHT	GRAY_LEVEL	Altitude-aware	Maximum absolute altitude difference between the node and any node in its subtree.
MEAN_GRAY_LEVEL	GRAY_LEVEL	Altitude-aware	Arithmetic mean of the image values over the full node support: sum(f(x), x in X) / card(X).
GRAY_LEVEL_VARIANCE	GRAY_LEVEL	Altitude-aware	Population variance of the image values over the full node support, with denominator card(X).

Gray-level attributes are good choices when the filter should depend on contrast, intensity, or altitude differences rather than pure geometry.

Size and Bounding-Box Attributes

Attribute	Groups	Contract	Use
AREA	SHAPE	Topology/support	Number of pixels in the full node support.
BOUNDING_BOX_WIDTH	SHAPE	Topology/support	Width, in columns, of the smallest axis-aligned bounding box enclosing the node support.
BOUNDING_BOX_HEIGHT	SHAPE	Topology/support	Height, in rows, of the smallest axis-aligned bounding box enclosing the node support.
DIAGONAL_LENGTH	SHAPE	Topology/support	Euclidean diagonal length of the bounding box, sqrt(width^2 + height^2).
RECTANGULARITY	SHAPE	Topology/support	Ratio AREA / (BOUNDING_BOX_WIDTH * BOUNDING_BOX_HEIGHT).
RATIO_WH	SHAPE	Topology/support	Bounding-box aspect ratio, max(width, height) / min(width, height) for non-degenerate boxes.
BOX_COLUMN_MIN	SHAPE	Topology/support	Minimum image column index covered by the node support.
BOX_COLUMN_MAX	SHAPE	Topology/support	Maximum image column index covered by the node support.
BOX_ROW_MIN	SHAPE	Topology/support	Minimum image row index covered by the node support.
BOX_ROW_MAX	SHAPE	Topology/support	Maximum image row index covered by the node support.

Distance-Transform Attributes

These attributes describe distances from support pixels to the foreground 4-connected contour. Contour pixels have distance zero. DIST_TRANSF uses the approximate field; DIST_TRANSF_EXACT uses exact Euclidean distances.

MAX_DIST is measured in pixels and equals the square root of MAX_SQUARED_DIST, which is measured in squared pixels. The same relation holds for MAX_DIST_EXACT and MAX_SQUARED_DIST_EXACT.

Approximate Distances

Attribute	Groups	Contract	Use
MAX_DIST	SHAPE, DIST_TRANSF	Topology/support	Maximum of the approximate distance field, in pixels.
MAX_SQUARED_DIST	DIST_TRANSF	Topology/support	Maximum approximate squared contour-distance cost, in squared pixels.
DIST_SQUARED_SUM	DIST_TRANSF	Topology/support	Sum of the approximate squared-distance field over the node support.
DIST_SQUARED_MEAN	DIST_TRANSF	Topology/support	Arithmetic mean of the approximate squared-distance field over the node support.
DIST_RMS	DIST_TRANSF	Topology/support	Square root of the mean approximate squared contour distance, in pixels.
DIST_SQUARED_VARIANCE	DIST_TRANSF	Topology/support	Population variance of the approximate squared-distance samples.
MAX_DIST_CENTER_ROW	DIST_TRANSF	Topology/support	Zero-based row of a support pixel attaining the approximate maximum; ties select the smallest row-major pixel.
MAX_DIST_CENTER_COLUMN	DIST_TRANSF	Topology/support	Zero-based column of the same deterministic approximate maximum-distance center.
MAX_DIST_PLATEAU_AREA	DIST_TRANSF	Topology/support	Number of support pixels attaining the approximate maximum.
MAX_DIST_PLATEAU_CENTROID_ROW	DIST_TRANSF	Topology/support	Arithmetic mean of row coordinates over the approximate maximum plateau.
MAX_DIST_PLATEAU_CENTROID_COLUMN	DIST_TRANSF	Topology/support	Arithmetic mean of column coordinates over the approximate maximum plateau.
DIST_SUM	DIST_TRANSF	Topology/support	Sum of approximate Euclidean contour distances over the node support, in accumulated pixel units.
DIST_MEAN	DIST_TRANSF	Topology/support	Arithmetic mean of Euclidean distances over the approximate field, in pixels.
DIST_VARIANCE	DIST_TRANSF	Topology/support	Population variance of approximate Euclidean contour distances, in squared pixels.
DIST_MEDIAN	DIST_TRANSF	Topology/support	Lower empirical median of the approximate Euclidean-distance field.
DIST_MODE	DIST_TRANSF	Topology/support	Most frequent approximate Euclidean distance; ties select the smallest distance.
DIST_Q25	DIST_TRANSF	Topology/support	Lower empirical 25th percentile of the approximate Euclidean-distance field.
DIST_Q75	DIST_TRANSF	Topology/support	Lower empirical 75th percentile of the approximate Euclidean-distance field.
DIST_Q90	DIST_TRANSF	Topology/support	Lower empirical 90th percentile of the approximate Euclidean-distance field.
DIST_ENTROPY	DIST_TRANSF	Topology/support	Shannon entropy in bits of the normalized approximate squared-distance histogram.
DIST_POSITIVE_AREA	DIST_TRANSF	Topology/support	Number of support pixels whose approximate squared-distance cost is positive.
DIST_LEVEL_COUNT	DIST_TRANSF	Topology/support	Number of represented approximate squared-distance levels, including zero.
DIST_WEIGHTED_CENTROID_ROW	DIST_TRANSF	Topology/support	Zero-based row centroid weighted by approximate Euclidean distance. All-zero fields use the ordinary support centroid.
DIST_WEIGHTED_CENTROID_COLUMN	DIST_TRANSF	Topology/support	Zero-based column centroid weighted by approximate Euclidean distance. All-zero fields use the ordinary support centroid.
DIST_WEIGHTED_CENTRAL_MOMENT_20	DIST_TRANSF	Topology/support	Unnormalized approximate distance-weighted central column moment.
DIST_WEIGHTED_CENTRAL_MOMENT_02	DIST_TRANSF	Topology/support	Unnormalized approximate distance-weighted central row moment.
DIST_WEIGHTED_CENTRAL_MOMENT_11	DIST_TRANSF	Topology/support	Unnormalized approximate distance-weighted mixed row-column moment.
DIST_WEIGHTED_AXIS_ORIENTATION	DIST_TRANSF	Topology/support	Principal axis of the approximate distance-weighted second moments, in degrees; isotropic fields return zero.
DIST_WEIGHTED_ECCENTRICITY	DIST_TRANSF	Topology/support	Major/minor eigenvalue ratio of the approximate distance-weighted second moments, capped at 1e6.

Exact Distances

Attribute	Groups	Contract	Use
MAX_DIST_EXACT	SHAPE, DIST_TRANSF_EXACT	Topology/support	Maximum exact Euclidean distance, in pixels, from a support pixel to the foreground 4-connected contour.
MAX_SQUARED_DIST_EXACT	DIST_TRANSF_EXACT	Topology/support	Maximum exact squared Euclidean contour distance, in squared pixels.
DIST_SQUARED_SUM_EXACT	DIST_TRANSF_EXACT	Topology/support	Sum of the exact squared Euclidean distance to the foreground A4 contour over every support pixel.
DIST_SQUARED_MEAN_EXACT	DIST_TRANSF_EXACT	Topology/support	Arithmetic mean of the exact squared contour-distance field over the node support.
DIST_RMS_EXACT	DIST_TRANSF_EXACT	Topology/support	Root mean square Euclidean contour distance, sqrt(mean(d^2)), in pixel units.
DIST_SQUARED_VARIANCE_EXACT	DIST_TRANSF_EXACT	Topology/support	Population variance of the exact squared contour-distance samples, mean((d^2)^2) - mean(d^2)^2, in fourth-power pixel units.
MAX_DIST_CENTER_ROW_EXACT	DIST_TRANSF_EXACT	Topology/support	Zero-based row of a support pixel attaining MAX_DIST_EXACT; ties select the smallest row-major pixel.
MAX_DIST_CENTER_COLUMN_EXACT	DIST_TRANSF_EXACT	Topology/support	Zero-based column of the same deterministic exact maximum-distance center.
MAX_DIST_PLATEAU_AREA_EXACT	DIST_TRANSF_EXACT	Topology/support	Number of support pixels attaining the exact maximum squared contour distance.
MAX_DIST_PLATEAU_CENTROID_ROW_EXACT	DIST_TRANSF_EXACT	Topology/support	Arithmetic mean of the row coordinates over the exact maximum-distance plateau.
MAX_DIST_PLATEAU_CENTROID_COLUMN_EXACT	DIST_TRANSF_EXACT	Topology/support	Arithmetic mean of the column coordinates over the exact maximum-distance plateau.
DIST_SUM_EXACT	DIST_TRANSF_EXACT	Topology/support	Sum of exact Euclidean contour distances over the support, in accumulated pixel units.
DIST_MEAN_EXACT	DIST_TRANSF_EXACT	Topology/support	Arithmetic mean of exact Euclidean contour distances, in pixels.
DIST_VARIANCE_EXACT	DIST_TRANSF_EXACT	Topology/support	Population variance of exact Euclidean contour distances, in squared pixels.
DIST_MEDIAN_EXACT	DIST_TRANSF_EXACT	Topology/support	Lower empirical median of the exact Euclidean-distance field.
DIST_MODE_EXACT	DIST_TRANSF_EXACT	Topology/support	Most frequent exact Euclidean distance; ties select the smallest distance.
DIST_Q25_EXACT	DIST_TRANSF_EXACT	Topology/support	Lower empirical 25th percentile of the exact Euclidean-distance field.
DIST_Q75_EXACT	DIST_TRANSF_EXACT	Topology/support	Lower empirical 75th percentile of the exact Euclidean-distance field.
DIST_Q90_EXACT	DIST_TRANSF_EXACT	Topology/support	Lower empirical 90th percentile of the exact Euclidean-distance field.
DIST_ENTROPY_EXACT	DIST_TRANSF_EXACT	Topology/support	Shannon entropy in bits of the normalized exact squared-distance histogram.
DIST_POSITIVE_AREA_EXACT	DIST_TRANSF_EXACT	Topology/support	Number of support pixels whose exact squared contour distance is positive.
DIST_LEVEL_COUNT_EXACT	DIST_TRANSF_EXACT	Topology/support	Number of represented exact squared-distance levels, including zero.
DIST_WEIGHTED_CENTROID_ROW_EXACT	DIST_TRANSF_EXACT	Topology/support	Zero-based row centroid weighted by exact Euclidean distance. All-zero fields use the ordinary support centroid.
DIST_WEIGHTED_CENTROID_COLUMN_EXACT	DIST_TRANSF_EXACT	Topology/support	Zero-based column centroid weighted by exact Euclidean distance. All-zero fields use the ordinary support centroid.
DIST_WEIGHTED_CENTRAL_MOMENT_20_EXACT	DIST_TRANSF_EXACT	Topology/support	Unnormalized exact distance-weighted central column moment.
DIST_WEIGHTED_CENTRAL_MOMENT_02_EXACT	DIST_TRANSF_EXACT	Topology/support	Unnormalized exact distance-weighted central row moment.
DIST_WEIGHTED_CENTRAL_MOMENT_11_EXACT	DIST_TRANSF_EXACT	Topology/support	Unnormalized exact distance-weighted mixed row-column moment.
DIST_WEIGHTED_AXIS_ORIENTATION_EXACT	DIST_TRANSF_EXACT	Topology/support	Principal axis of the exact distance-weighted second moments, in degrees; isotropic fields return zero.
DIST_WEIGHTED_ECCENTRICITY_EXACT	DIST_TRANSF_EXACT	Topology/support	Major/minor eigenvalue ratio of the exact distance-weighted second moments, capped at 1e6.

Filled-Shape Attributes

These descriptors measure the region enclosed by the unique external boundary, including its holes. Each node must have exactly one external boundary. HOLE_AREA_FRACTION and FILLED_CENTROID_DISPLACEMENT_NORMALIZED compare that filled region with the original support.

Attribute	Groups	Contract	Use
FILLED_AREA	FILLED_SHAPE	Topology/support	Number of pixels enclosed by the unique external boundary, including holes; requires exactly one external boundary per live node.
FILLED_CENTROID_ROW	FILLED_SHAPE	Topology/support	Zero-based mean row of the filled region pixels.
FILLED_CENTROID_COLUMN	FILLED_SHAPE	Topology/support	Zero-based mean column of the filled region pixels.
FILLED_LENGTH_MAJOR_AXIS	FILLED_SHAPE	Topology/support	Major-axis length proxy of the filled region, in pixels; follows LENGTH_MAJOR_AXIS.
FILLED_LENGTH_MINOR_AXIS	FILLED_SHAPE	Topology/support	Minor-axis length proxy of the filled region, in pixels; follows LENGTH_MINOR_AXIS.
FILLED_AXIS_ORIENTATION	FILLED_SHAPE	Topology/support	Non-negative principal column-axis orientation in degrees; isotropic regions return 0.
FILLED_ECCENTRICITY	FILLED_SHAPE	Topology/support	Major/minor eigenvalue ratio of the filled region moments, capped at 1e6; one pixel returns 1.
FILLED_INERTIA	FILLED_SHAPE	Topology/support	Sum of the filled region second central moments divided by squared filled area.
HOLE_AREA_FRACTION	FILLED_SHAPE	Topology/support	Fraction of filled area occupied by holes: 1 - AREA / FILLED_AREA.
FILLED_CENTROID_DISPLACEMENT_NORMALIZED	FILLED_SHAPE	Topology/support	Euclidean distance between support and filled-region centroids divided by the square root of filled area.
FILLED_COMPACTNESS	FILLED_SHAPE	Topology/support	FILLED_AREA / (2*pi*(mu20 + mu02)), using filled-region second central moments; zero dispersion returns 0.
FILLED_CIRCULARITY	FILLED_SHAPE	Topology/support	Minor/major eigenvalue ratio of the filled-region second moments; one pixel returns 1 and a nontrivial line returns 0.

Moment Attributes

Attribute	Groups	Contract	Use
CENTRAL_MOMENT_20	MOMENTS, SHAPE	Topology/support	Second-order central moment mu20 around the support centroid.
CENTRAL_MOMENT_02	MOMENTS, SHAPE	Topology/support	Second-order central moment mu02 around the support centroid.
CENTRAL_MOMENT_11	MOMENTS, SHAPE	Topology/support	Mixed second-order central moment mu11.
CENTRAL_MOMENT_30	MOMENTS, SHAPE	Topology/support	Third-order central moment mu30.
CENTRAL_MOMENT_03	MOMENTS, SHAPE	Topology/support	Third-order central moment mu03.
CENTRAL_MOMENT_21	MOMENTS, SHAPE	Topology/support	Mixed third-order central moment mu21.
CENTRAL_MOMENT_12	MOMENTS, SHAPE	Topology/support	Mixed third-order central moment mu12.
HU_MOMENT_1	MOMENTS, SHAPE	Topology/support	First Hu moment invariant computed from normalized central moments.
HU_MOMENT_2	MOMENTS, SHAPE	Topology/support	Second Hu moment invariant.
HU_MOMENT_3	MOMENTS, SHAPE	Topology/support	Third Hu moment invariant.
HU_MOMENT_4	MOMENTS, SHAPE	Topology/support	Fourth Hu moment invariant.
HU_MOMENT_5	MOMENTS, SHAPE	Topology/support	Fifth Hu moment invariant.
HU_MOMENT_6	MOMENTS, SHAPE	Topology/support	Sixth Hu moment invariant.
HU_MOMENT_7	MOMENTS, SHAPE	Topology/support	Seventh Hu moment invariant.
INERTIA	MOMENTS, SHAPE	Topology/support	Sum of normalized second-order central moments, mu20 / area^2 + mu02 / area^2.
COMPACTNESS	MOMENTS, SHAPE	Topology/support	Area normalized by second-order dispersion, (1 / (2*pi)) * area / (mu20 + mu02) when the denominator is positive.
ECCENTRICITY	MOMENTS, SHAPE	Topology/support	Ratio of the largest to smallest eigenvalue of the second-moment matrix.
LENGTH_MAJOR_AXIS	MOMENTS, SHAPE	Topology/support	Length proxy for the major axis of the equivalent second-moment ellipse, derived from the largest inertia eigenvalue and area.
LENGTH_MINOR_AXIS	MOMENTS, SHAPE	Topology/support	Length proxy for the minor axis of the equivalent second-moment ellipse, derived from the smallest inertia eigenvalue and area.
AXIS_ORIENTATION	MOMENTS, SHAPE	Topology/support	Principal-axis orientation in degrees, 0.5 * atan2(2*mu11, mu20 - mu02), normalized to a non-negative angle.
CIRCULARITY	MOMENTS, SHAPE	Topology/support	Ratio lambda2 / lambda1 of second-moment eigenvalues.

Moment descriptors are useful when area alone is too coarse and the filter should react to shape geometry.

Boundary and Bitquad Attributes

Attribute	Groups	Contract	Use
BITQUAD_AREA	BOUNDARY, SHAPE	Topology/support	Duda-style sub-pixel area estimator derived from aggregated 2x2 bitquad pattern counts.
BITQUAD_NUMBER_EULER	BOUNDARY, SHAPE	Topology/support	Euler-characteristic estimate from bitquad counts under the tree connectivity.
BITQUAD_NUMBER_HOLES	BOUNDARY, SHAPE	Topology/support	Number of holes inferred from the bitquad Euler characteristic for a single connected support.
BITQUAD_PERIMETER	BOUNDARY, SHAPE	Topology/support	Discrete boundary-length estimate from bitquad edge-contributing patterns.
BITQUAD_PERIMETER_CONTINUOUS	BOUNDARY, SHAPE	Topology/support	Smoothed continuous perimeter estimate from bitquad counters, using weighted transitions across local 2x2 configurations.
BITQUAD_CIRCULARITY	BOUNDARY, SHAPE	Topology/support	Bitquad compactness measure (4*pi*BITQUAD_AREA) / BITQUAD_PERIMETER_CONTINUOUS^2.
BITQUAD_PERIMETER_AVERAGE	BOUNDARY, SHAPE	Topology/support	Average continuous perimeter per connected component, computed from bitquad perimeter and Euler-count estimates.
BITQUAD_LENGTH_AVERAGE	BOUNDARY, SHAPE	Topology/support	Average longitudinal extent proxy, derived as half of the average continuous perimeter.
BITQUAD_WIDTH_AVERAGE	BOUNDARY, SHAPE	Topology/support	Average transverse extent proxy, computed as 2 * BITQUAD_AREA / BITQUAD_PERIMETER_CONTINUOUS with a zero fallback when the continuous perimeter is degenerate.
CONTOUR_PIXELS	BOUNDARY, SHAPE	Topology/support	Number of support pixels that touch the 4-neighbor complement by at least one side.
CONTOUR_PERIMETER	BOUNDARY, SHAPE	Topology/support	Total number of exposed 4-neighbor sides over the support.
CONTOUR_SIDE_NORTH	BOUNDARY, SHAPE	Topology/support	Number of exposed north-facing sides over support pixels.
CONTOUR_SIDE_WEST	BOUNDARY, SHAPE	Topology/support	Number of exposed west-facing sides over support pixels.
CONTOUR_SIDE_EAST	BOUNDARY, SHAPE	Topology/support	Number of exposed east-facing sides over support pixels.
CONTOUR_SIDE_SOUTH	BOUNDARY, SHAPE	Topology/support	Number of exposed south-facing sides over support pixels.

Boundary attributes are useful when perimeter, holes, or contour exposure are more informative than raw area.

Tree-Topology Attributes

Attribute	Groups	Contract	Use
SUBTREE_HEIGHT	TREE_TOPOLOGY	Tree topology	Longest child-edge path from the node to any leaf in its subtree.
DEPTH_NODE	TREE_TOPOLOGY	Tree topology	Number of parent-edge steps from the node to the root.
IS_LEAF_NODE	TREE_TOPOLOGY	Tree topology	Boolean scalar encoded as 1 when the node has no children and 0 otherwise.
IS_ROOT_NODE	TREE_TOPOLOGY	Tree topology	Boolean scalar encoded as 1 for the tree root and 0 for all other nodes.
NUM_CHILDREN_NODE	TREE_TOPOLOGY	Tree topology	Number of direct child nodes.
NUM_SIBLINGS_NODE	TREE_TOPOLOGY	Tree topology	Number of other nodes sharing the same parent.
NUM_DESCENDANTS_NODE	TREE_TOPOLOGY	Tree topology	Number of internal tree nodes strictly below this node in its subtree.
NUM_LEAF_DESCENDANTS_NODE	TREE_TOPOLOGY	Tree topology	Number of leaf nodes in the node subtree.
LEAF_RATIO_NODE	TREE_TOPOLOGY	Tree topology	Ratio of leaf descendants to subtree size, leaf_descendants / (descendants + 1).
BALANCE_NODE	TREE_TOPOLOGY	Tree topology	Difference between maximum and minimum child-subtree heights.
AVG_CHILD_HEIGHT_NODE	TREE_TOPOLOGY	Tree topology	Average height of direct child subtrees.

Tree-topology attributes ignore image coordinates and gray levels. They are useful when the hierarchy shape itself carries the signal.

Choosing Attributes

Start small:

• use AREA for size filtering;

• add GRAY_LEVEL_HEIGHT or VOLUME when contrast matters;

• add COMPACTNESS, ECCENTRICITY, or RATIO_WH when component shape matters;

• add TREE_TOPOLOGY attributes when hierarchy position matters;

• use BOUNDARY attributes for contour- or hole-sensitive filters;

• use DIST_TRANSF or DIST_TRANSF_EXACT for interior-distance descriptors;

• use FILLED_SHAPE to describe enclosed geometry and the effect of holes.

For CFP layers, each scalar attribute becomes an input feature for the scoring model after group expansion.

filter_spec = {
    "name": "area_contrast_shape",
    "tree_type": "max-tree",
    "attributes": [
        morphology.AttributeType.AREA,
        morphology.AttributeType.GRAY_LEVEL_HEIGHT,
        morphology.AttributeType.COMPACTNESS,
    ],
}