mmcfilters 0.1.25

Library for connected image filtering based on morphological trees

MorphologicalAttributeFilters

MorphologicalAttributeFilters is a C++/Python library for connected image filtering based on morphological trees (component trees and trees of shapes). The code provides high-performance implementations together with Python bindings for interactive experimentation.

Key features

• Construction of morphological trees (component tree, tree of shapes) with different connectivities.
• Incremental computation of geometric, bit-quads, topological, stats and other attributes.
• Attribute-based filters (direct and subtractive rules).
• Utilities for extinction values and MSER.
• Pybind11 bindings that expose the high-level operations to Python.

Installation

pip install mmcfilters

Quick example

import numpy as np
import mmcfilters
Type = mmcfilters.Attribute.Type

img = np.random.randint(0, 255, size=(128, 128), dtype=np.uint8)

maxtree = mmcfilters.MorphologicalTree(img, True, 1.5)
#mintree = mmcfilters.MorphologicalTree(img, False, 1.5)
#tos = mmcfilters.MorphologicalTree(img) 

filter = mmcfilters.AttributeFilters(maxtree)
area = mmcfilters.Attribute.computeSingleAttribute(maxtree, Type.AREA)

img_filtered = filter.filteringDirectRule(area > 50)