pixel-canon 0.1.1

The canon for pixel data topology. A cross-language specification to define the logical layout of images.

English | Русский

---

Pixel-Canon

The canon for pixel data topology. A cross-language specification to define the logical layout of images (axis order, orientation, and memory order).

Pixel-Canon is a cross-language project aimed at solving a common and frustrating problem in computer vision, image processing, and machine learning: the ambiguity of image data layouts. When you receive an N-dimensional array, what do the axes mean? Is it (Height, Width, Channels) or (Channels, Height, Width)? Does the Y-axis point up or down? Are the data stored in C-style (row-major) or Fortran-style (column-major) order?

This project provides a simple, declarative specification and a set of tools to describe this information explicitly, eliminating guesswork and making data pipelines more robust and reliable.

Core Concepts

The specification is built on a few core components:

• ImageAxis: The name of an axis (e.g., WIDTH, HEIGHT, CHANNELS).
• AxisDirection: The orientation of an axis (e.g., UP, DOWN, LEFT, RIGHT).
• MemoryOrder: The physical layout in memory (ROW_MAJOR or COLUMN_MAJOR).
• ChannelFormat: The semantic order of channels (RGB, BGR, RGBA).
• ImageLayout: A composite object that combines all of the above to provide a complete, unambiguous description.

The Specification

The formal specification, which all language-specific implementations must adhere to, can be found here:

• Specification v1.0

Implementations

• Python (In Progress)
• Other languages to be added.

License

This project is licensed under the MIT License. See the LICENSE file for details.