MorphoTreeViz – Morphological Trees Visualization

Interactive visualization of morphological trees.

Quick Start (import mtviz as viz)

import mtviz as viz

class Node:
    def __init__(self, id, level, area, children=None):
        self.id = id
        self.level = level
        self.area = area
        self.children = children or []

root = Node(1, 0, 10, [
    Node(2, 1, 5, [Node(4, 2, 3), Node(5, 2, 2)]),
    Node(3, 1, 5, [Node(6, 2, 1)])
])

pt = viz.PrintTree(lambda n: n.children, lambda n: f"{n.id}: {n.level}: {n.area}")
pt(root)

Color (optional)

• Pass ANSI codes in color (e.g., "\x1b[40m\x1b[37m" for black background + white text).
• If you use colorama in your project, you can pass Back.BLACK + Fore.WHITE.

Interactive Tree (viz.makePlotTree)

Use Bokeh to visualize and interact with the tree: hover tooltips, click actions, flood tools, and size/zoom controls.

import mtviz as viz
from bokeh.io import show, output_notebook

output_notebook()  # inline in Jupyter; use show(p) to open in browser

class Node:
    def __init__(self, id, level, area, children=None):
        self.id = id
        self.level = level
        self.area = area
        self.children = children or []

root = Node(1, 0, 10, [
    Node(2, 1, 5, [Node(4, 2, 3), Node(5, 2, 2)]),
    Node(3, 1, 5, [Node(6, 2, 1)])
])

p = viz.makePlotTree(
    root,
    lambda n: n.children,
    {
      "id:":      lambda n: n.id,
      "level:":   "level",
      "area:":    lambda n: n.area,
      "repCNPs":  lambda n: getattr(n, 'repCnps', []),
    },
    lambda n: n.id,
    node_size=18,
    flood_use=None,
)
show(p.panel)

Image Overlay + Flood

You can compose a figure with a grayscale image and an RGBA overlay, then link DrawTree’s flood tools.

import mtviz as viz
from bokeh.layouts import row, column

img = ...  # 2D numpy array (grayscale)
bundle_img = viz.makePlotImage(img)
p_img = bundle_img.panel
img_source = bundle_img.source
alpha_slider = bundle_img.controls[0]

p_tree = viz.makePlotTree(
  root,
  lambda n: n.children,
  { "id:": lambda n: n.id, "repCNPs": lambda n: getattr(n, 'repCnps', []) },
  lambda n: n.id,
  image_source=img_source,
  alpha_slider=alpha_slider,
  flood_use='floodThreshold',
  flood_params={'polarity': 255},
)
show(row(p_tree, column(alpha_slider, p_img)))

Pixel selection

makePlotImage exposes the last clicked pixel through bundle.selection_source. The source contains one row with row, column, index, and value, and is updated by CustomJS in the browser.

bundle = viz.makePlotImage(
    img,
    selected_pixel=(20, 35),
    title="Click a pixel",
)

selection = bundle.selection_source
viz.show(bundle)

The selection source can drive other Bokeh models without a Bokeh server. For the common morphological-tree use case, makePlotImageTreeInspector links one image selector directly to the node support in several trees:

inspector = viz.makePlotImageTreeInspector(
    img,
    {
        "Tree of Shapes": tree_of_shapes,
        "Residual Tree": residual_tree,
    },
    initial_pixel=(20, 35),
    width=520,
    support_width=390,
)
viz.show(inspector)

The tree objects must expose num_nodes, smallest_node_map, and parent(node). When available, node altitude, residue, proper-part size, and branch length are included in the live readout. Clicking the input image updates every support view immediately in standalone HTML and Jupyter output.