Documentation

This is machine translation

Translated by Microsoft
Mouseover text to see original. Click the button below to return to the English verison of the page.

Note: This page has been translated by MathWorks. Please click here
To view all translated materals including this page, select Japan from the country navigator on the bottom of this page.

undistortFisheyePoints

Correct point coordinates for fisheye lens distortion

Syntax

undistortedPoints = undistortFisheyePoints(points,intrinsics)
undistortedPoints = undistortFisheyePoints(___,scaleFactor)
[___,camIntrinsics] = undistortFisheyePoints(___)
[___,reprojectionErrors] = undistortFisheyePoints(___)

Description

example

undistortedPoints = undistortFisheyePoints(points,intrinsics) returns point coordinates corrected for fisheye lens distortion.

undistortedPoints = undistortFisheyePoints(___,scaleFactor) returns corrected point coordinates using the scaleFactor and the previous inputs.

[___,camIntrinsics] = undistortFisheyePoints(___) also returns a cameraIntrinsics object, which corresponds to a virtual perspective camera that produces undistorted points.

[___,reprojectionErrors] = undistortFisheyePoints(___) also returns reprojectionErrors used to evaluate the accuracy of undistorted points. The function computes the reprojection errors by applying distortion to the points, and taking the distances between the result and the corresponding input points.

Examples

collapse all

Undistort and translate checkerboard points detected in a calibration image, and then display the results.

Create an imageDatastore object containing checkerboard calibration images.

images = imageDatastore(fullfile(toolboxdir('vision'),'visiondata',...
   'calibration','gopro'));
imageFileNames = images.Files;

Detect the calibration pattern from the images.

[imagePoints,boardSize] = detectCheckerboardPoints(imageFileNames);

Generate world coordinates for the corners of the checkerboard squares.

squareSize = 29; % millimeters
worldPoints = generateCheckerboardPoints(boardSize,squareSize);

Estimate the fisheye parameters from the image and world points. Get the image size from the first image.

I = readimage(images,10);
imageSize = [size(I,1) size(I,2)];
params = estimateFisheyeParameters(imagePoints,worldPoints,imageSize);

In the first image, detect the checkerboard points.

points = detectCheckerboardPoints(I);

Undistort the points and image.

[undistortedPoints,intrinsics1] = undistortFisheyePoints(points,params.Intrinsics);
[J, intrinsics2] = undistortFisheyeImage(I,params.Intrinsics,'OutputView','full');

Translate the undistorted points.

newOrigin = intrinsics2.PrincipalPoint - intrinsics1.PrincipalPoint;
undistortedPoints = [undistortedPoints(:,1) + newOrigin(1), ...
                    undistortedPoints(:,2) + newOrigin(2)];

Display the results.

figure
imshow(I)
hold on
plot(points(:,1),points(:,2),'r*-')
title('Detected Points')
hold off

figure
imshow(J)
hold on
plot(undistortedPoints(:, 1),undistortedPoints(:, 2),'g*-')
title('Undistorted Points')
hold off

Input Arguments

collapse all

Input points, specified as an M-by-2 matrix of M [x y] coordinates.

Fisheye intrinsic camera parameters, specified as a fisheyeIntrinsics object.

Scale factor for points, specified as a scalar or an [sx sy] vector. Specify a vector to scale the x and y axes individually. Increase the scale to zoom in the perspective of the camera view.

Output Arguments

collapse all

Undistorted points, returned as an M-by-2 matrix of M number of [x y] coordinates. If points is double, then undistortedPoints is double. Otherwise, undistortedPoints is single.

Data Types: single | double

Undistorted intrinsics of a virtual camera, returned as a cameraIntrinsics object. These intrinsics are for a camera that has a perspective that produces the undistorted image.

Reprojection errors, returned as an M-by-1 vector. The function computes the reprojection errors by applying distortion to the undistorted points and taking the distances between the results and the corresponding input points. Errors are expressed in pixels.

Introduced in R2017b