Pywayne Camera Model

This module wraps the camera_models C++ library via pybind11, providing a Python interface for camera operations.

Quick Start

from pywayne.cv.camera_model import CameraModel
from pywayne.cv.tools import write_cv_yaml
import numpy as np

# Create camera model
camera = CameraModel()

# Load from YAML file
camera.load_from_yaml('camera_config.yaml')

# Access properties
print(f"Model: {camera.model_type}")
print(f"Size: {camera.image_width}x{camera.image_height}")

Loading from YAML

from pathlib import Path

# Sample YAML configuration
yaml_data = {
    "model_type": "PINHOLE",
    "camera_name": "my_camera",
    "image_width": 1280,
    "image_height": 720,
    "distortion_parameters": {"k1": 0.0, "k2": 0.0},
    "projection_parameters": {"fx": 600.0, "fy": 600.0, "cx": 640.0, "cy": 360.0}
}

# Write YAML file
write_cv_yaml('camera_config.yaml', yaml_data)

# Load model
camera.load_from_yaml('camera_config.yaml')

Supported Camera Models

Projection Operations

lift_projective()

Lifts a 2D image point to a 3D projective ray:

# Input can be tuple, list, or numpy array
ray_3d = camera.lift_projective([u, v])  # Returns np.ndarray (x, y, z)

space_to_plane()

Projects a 3D point onto the 2D image plane:

# Input can be tuple, list, or numpy array
uv = camera.space_to_plane([x, y, z])  # Returns np.ndarray (u, v)

Properties

Parameters Dictionary

Export all camera parameters as a dictionary:

params = camera.get_parameters_as_dict()
print(params)

Includes model-specific parameters:

• Pinhole: k1, k2, p1, p2, fx, fy, cx, cy
• Pinhole Full: k1-k6, p1-p2, fx, fy, cx, cy
• CATA: xi, k1-k2, p1-p2, gamma1-2, u0, v0
• Equidistant: k2-k5, mu, mv, u0, v0
• OCAM: C, D, E, center_x, center_y, poly, inv_poly

Requirements

• camera_models - C++ library (auto-downloaded via gettool if missing)
• numpy - Array operations
• pywayne.cv.tools.write_cv_yaml - For writing YAML files

Notes

• Library is automatically checked and downloaded via gettool if not found
• Supports both tuple/list and numpy array inputs for projection methods
• Output from projection methods is always np.ndarray with dtype=np.float64