CalibratedCamera - GTSAM Docs

CalibratedCamera is the variable type to use when camera calibration is known exactly and the only unknowns are the camera’s position and attitude. It is a six-degree-of-freedom pinhole camera that operates directly in normalized image coordinates.

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

Creating a camera¶

The camera stores a Pose3. The identity pose places the camera at the world origin looking along its positive z-axis.

pose = gtsam.Pose3(gtsam.Rot3.Yaw(0.1), np.array([1.0, 0.0, 0.0]))
camera = gtsam.CalibratedCamera(pose)
print("camera center:", camera.pose().translation())

Projection and backprojection¶

project(point) returns (x/z, y/z) in the camera frame. backproject(measurement, depth) follows that ray to the requested camera-frame depth.

point = np.array([2.0, 0.5, 4.0])
measurement = camera.project(point)
depth = camera.pose().transformTo(point)[2]
recovered = camera.backproject(measurement, depth)
print("normalized measurement:", measurement)
np.testing.assert_allclose(recovered, point, atol=1e-9)

Geometry and manifold operations¶

range() computes distance to a point, pose, or camera. retract() and localCoordinates() expose the six-dimensional camera-pose manifold. Level() is a convenient factory for lifting a planar pose to a level camera at a specified height.

level_camera = gtsam.CalibratedCamera.Level(gtsam.Pose2(2.0, 3.0, 0.2), 1.5)
print("level camera center:", level_camera.pose().translation())
print("range to point:", camera.range(point))

Source¶

CalibratedCamera.h

AI assistance caveat¶

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