i2g 0.2.0

Convert grayscale images into NetworkX graph structures

i2g — Image to Graph Converter

Convert grayscale images into graph structures using NetworkX. Each pixel becomes a node connected to its neighbors (4- or 8-connectivity), with pixel intensity stored as a node attribute.

---

Installation

pip install i2g

Or install from source in editable mode:

git clone https://github.com/DIM-Corp/i2g.git
cd i2g
pip install -e .[dev]

Requires Python 3.10+. Dependencies: numpy, Pillow, networkx.

---

Usage

from i2g import ImageGraphConverter

converter = ImageGraphConverter("my_image.png", connectivity="8")
graph, img_array = converter.convert()

shape = converter.shape()          # (height, width)
num_nodes, num_edges = converter.info()

print(f"Image shape: {shape}")
print(f"Graph has {num_nodes} nodes and {num_edges} edges.")

Node attributes

Each node is keyed by (row, col) and carries two attributes:

Attribute	Type	Description
intensity	int	Grayscale pixel value (0–255)
pos	tuple[int, int]	(col, -row) — ready for matplotlib/networkx plotting

Connectivity

Value	Neighbors
"4"	Up, down, left, right
"8" (default)	Cardinal + diagonal (8 neighbors)

Error handling

Situation	Exception raised
Invalid connectivity value	ValueError at construction time
Image file not found	FileNotFoundError from convert()
File exists but is not a valid image	OSError from convert()
Image exceeds 10 million pixels	ValueError from convert()
shape() or info() called before convert()	RuntimeError

---

Running tests

pytest

---

License

MIT License — see LICENSE for details.

---

Contact

[e-mail] · [e-mail]