Class Camera

Defined in File camera.hpp

Class Documentation

class Camera

This class implements a camera models, made up of intrinsic and extrinsic parameters (lib3d::Intrinsics & lib3d::Extrinsics).

The intrinsic parameters allow to model a perspective or parallel camera with or without distortion. The distortion models are implemented by OpenCV and supports radial, tangential and prism distortion. For more information see OpenCV-Website.

Example

Initialization of a perspecitve camera with no distortion (pinhole camera) and no rotation or translation use:

lib3d::Camera pinholeCamera(lib3d::Intrinsics(
                             cv::Size(1920, 1080),          // image size
                             cv::Point2d(1273.38, 1231.75), // focal length
                             cv::Point2d(960,540)           // principal point
                            ));

Public Functions

inline explicit Camera()

Default (zero initialization) constructor.

This will initialize the camera with no intrinsic and extrinsic parameters.

inline explicit Camera(const Intrinsics &intr)

Initialize camera with given intrinsic parameters.

Parameters:: intr – [in] Intrinsic camera parameters.

inline explicit Camera(const Intrinsics &intr, const Extrinsics &extr)

Initialize camera with given intrinsic and extrinsic parameters.

Parameters:

intr – [in] Intrinsic camera parameters.
extr – [in] Extrinsic camera parameters.

inline Camera(const Camera &rhs): Copy constructor.

inline Camera(Camera &&rhs): Move constructor.

inline Camera &operator=(const Camera &rhs): Copy assignment operator.

inline Camera &operator=(Camera &&rhs): Move assignment operator.

inline bool operator==(const Camera &rhs)

Comparison operator.

Returns:: True, if the intrinsic, distortion and extrinsic parameters are equal.

inline bool operator!=(const Camera &rhs)

Comparison operator.

Returns:: True, if the intrinsic, distortion and extrinsic parameters are NOT equal.

inline cv::Point2i projectWorld2Pixel(cv::Point3f worldPnt) const

Method to project a 3D world point into 2D pixel coordinates.

This will first transform the world point into local camera coordinates by applying extrinsics.getRTMatrix(lib3d::Extrinsics::REF_2_LOCAL) to the world point. Then it will perform a projection by getting the specific projeciton model from the instrinsic parameters. This will yield a point in image coordinates on the focal plance, centered in the principle point. This image point is then projected into pixel coordinates according to the calibration matrix \(\mathrm{K}\).

This function will internally use the function projectPoints from OpenCV as this also considers the distortion parameters.

Also see: https://docs.opencv.org/master/d9/d0c/group__calib3d.html

inline cv::Point2i projectWorld2Pixel(float worldPnt_x, float worldPnt_y, float worldPnt_z) const: This is an overloaded member function, provided for convenience. It differs from the above function only in what argument(s) it accepts.

inline cv::Point2i projectLocal2Pixel(cv::Point3f localPnt) const

Method to project a 3D point given in local coordinates of the camera into 2D pixel coordinates.

This is similar to projectWorld2Pixel, only that it omits the transformation with extrinsics.getRTMatrix(lib3d::Extrinsics::REF_2_LOCAL). In this it es assumed that the local point is given in the coordinate system of the camera with the x-axis pointing to the right, the y-axis to the bottom and the z-axis to the viewing direction of the camera.

This function will internally use the function projectPoints from OpenCV as this also considers the distortion parameters.

Also see: https://docs.opencv.org/master/d9/d0c/group__calib3d.html

inline cv::Point2i projectLocal2Pixel(float localPnt_x, float localPnt_y, float localPnt_z) const: This is an overloaded member function, provided for convenience. It differs from the above function only in what argument(s) it accepts.

Public Members

Intrinsics intrinsics: Intrinsic camera parameters.

Extrinsics extrinsics: Extrinsic camera parameters.