The answer to this question depends on undistortImage function's Name-Value argument OutputView. If it is not used and left to use its default value of OutputView="same", then the new camera matrix will be the same as the original camera matrix in the exported cameraParameters.
If the OutputView is set to "full" or "valid", then the principal point of the original camera matrix must be translated to form the new camera matrix. In short, you could do the following to account for all these 3 cases:
% Undistort the image specifying the desired OutputView.
[J, newOrigin] = undistortImage(I, cameraParams, OutputView=view)
% Translate the principal point. newOrigin is zero if OutputView="same"
newPrincipalPoint = cameraParams.PrincipalPoint - newOrigin;
newImageSize = size(J,1:2);
% Create a new cameraIntrinsics object which contains the new camera matrix.
newIntrinsics = cameraIntrinsics(cameraParams.FocalLength, newPrincipalPoint, ...
newImageSize, Skew=cameraParams.Skew);
