Cal3 - GTSAM Docs

The most common calibration object is affine and can be represented simply by a matrix K. Cal3 stores its five linear pinhole intrinsics—horizontal and vertical focal lengths, skew, and principal point—and is the common base interface for GTSAM’s richer camera calibration models.

Authors: Frank Dellaert, et al. (see THANKS for the full author list)

See LICENSE for the license information

try:
    import google.colab
    %pip install --quiet gtsam-develop
except ImportError:
    pass

import gtsam
import numpy as np

Initialization¶

Construct from five scalars in (fx, fy, skew, px, py) order or from a five-element vector.

calibration = gtsam.Cal3(800.0, 790.0, 0.2, 320.0, 240.0)
from_vector = gtsam.Cal3(np.array([800.0, 790.0, 0.2, 320.0, 240.0]))
assert calibration.equals(from_vector, 1e-12)

Properties¶

fx(), fy(), skew(), px(), and py() expose the five parameters. principalPoint() and aspectRatio() provide common derived quantities, and vector() returns the constructor order.

print("parameters:", calibration.vector())
print("principal point:", calibration.principalPoint())
print("aspect ratio fy/fx:", calibration.aspectRatio())

Matrix form¶

K() returns the standard upper-triangular intrinsic matrix and inverse() returns its inverse. These are useful when converting between normalized homogeneous coordinates and pixel coordinates.

K = calibration.K()
K_inverse = calibration.inverse()
print(K)
np.testing.assert_allclose(K_inverse @ K, np.eye(3), atol=1e-12)

normalized_h = np.array([0.1, -0.05, 1.0])
pixel_h = K @ normalized_h
print("pixel coordinates:", pixel_h[:2] / pixel_h[2])

Base-class role¶

Cal3 itself models only the linear intrinsic matrix and is not exposed as an optimizable manifold. Concrete classes such as Cal3_S2, Cal3DS2, and Cal3Fisheye add projection and manifold operations.

Source¶

Cal3.h

AI assistance caveat¶

AI was used to help draft this documentation, and inaccuracies could be present.