You are here: Start » AVL.NET » AVL.CalibrateCamera Method

AVL.CalibrateCamera Method

Finds the camera intrinsic and extrinsic parameters from the input arrays of image and real-world coordinates.

Namespace:	AvlNet
Assembly:	AVL.NET.dll

Syntax

public static void CalibrateCamera(
	AvlNet.Point3D[][] inObjectPoints,
	AvlNet.Point2D[][] inImagePoints,
	int inImageWidth,
	int inImageHeight,
	bool inFixPrincipalPoint,
	bool inFixAspectRatio,
	bool inZeroTangentDist,
	AvlNet.Matrix inInitialCameraMatrix,
	bool inUseRationalModel,
	bool inUseThinPrismModel,
	out AvlNet.Matrix outCameraMatrix,
	out AvlNet.LensDistortion outDistortion,
	out float[] outUnitScales,
	out AvlNet.Position3D[] outCameraPositions,
	out float outError
)

Parameters

Name	Type	Range	Default	Description
inObjectPoints	AvlNet.Point3D			Array, for each view: array of 3D points of the calibration pattern.
inImagePoints	AvlNet.Point2D			Array, for each view: array of corresponding 2D points in the picture.
inImageWidth	int	<1, INF>	640	Image width, only used to initialize camera matrix, not used when inInitialCameraMatrix present. Default value: 640.
inImageHeight	int	<1, INF>	480	Image height, only used to initialize camera matrix, not used when inInitialCameraMatrix present. Default value: 480.
inFixPrincipalPoint	bool			The principal point is not changed during the global optimization. It stays in the center of the image, or as in the initial camera matrix.
inFixAspectRatio	bool			The functions considers only fy as a free parameter. The ratio fx/fy stays 1, or as in the initial camera matrix.
inZeroTangentDist	bool			Tangential distortion coefficients (p_1, p_2) are assumed zero.
inInitialCameraMatrix	AvlNet.Matrix			Optional initial camera matrix, required for inFixPrincipalPoint and inFixAspectRatio. Default value: atl::NIL, or null.
inUseRationalModel	bool
inUseThinPrismModel	bool
outCameraMatrix	AvlNet.Matrix
outDistortion	AvlNet.LensDistortion
outUnitScales	float			Image scale: px / mm.
outCameraPositions	AvlNet.Position3D			Positions of cameras in world coordinate system for the views
outError	float			Final re-projection error value

Description

The coordinates of 3D object points and their corresponding 2D projections in each view must be specified. That may be achieved by using an object with a known geometry and easily detectable feature points. Such an object is called a calibration rig or calibration pattern. One example is a chessboard (see DetectChessboardGrid).

When camera matrix and distortion coefficients are calculated, one can create undistortion maps using CreateUndistortionMap function. These maps can be applied to images using RemapImage.

Remarks

Currently, automatic initialization of intrinsic parameters (when inInitialCameraMatrix is not specified) is only implemented for planar calibration patterns (where Z-coordinates of the object points must be all zeros). 3D calibration rigs can also be used as long as initial camera matrix is provided.

Providing inImageWidth and inImageHeight is important, when inInitialCameraMatrix is not specified. This is used to determine the principal point (part of the initial camera matrix). If this point is assumed to be far from the actual image center, false tangential distortion can be determined by the calibration algorithm. This manifests itself especially when the number of point correspondences is relatively small.

Errors