gusucode.com > vision工具箱matlab源码程序 > vision/rectifyStereoImages.m
% rectifyStereoImages Rectifies a pair of stereo images.
% [J1, J2] = rectifyStereoImages(I1, I2, stereoParams) undistorts and rectifies
% I1 and I2, a pair of truecolor or grayscale stereo images. stereoParams
% is a stereoParameters object containing the parameters of the
% stereo camera system. J1 and J2 are the rectified images.
%
% [J1, J2] = rectifyStereoImages(I1, I2, tform1, tform2) rectifies I1 and
% I2, by applying projective transformations tform1 and tform2 returned
% by the estimateUncalibratedRectification function.
%
% [...] = rectifyStereoImages(..., interp) specifies interpolation
% method to use. interp can be one of the strings 'nearest',
% 'linear', or 'cubic'. The default value for interp is 'linear'.
%
% [...] = rectifyStereoImages(..., Name, Value) specifies additional
% name-value pairs described below:
%
% 'OutputView' Determines the size of the rectified images J1 and J2.
% Possible values are:
% 'full' - J1 and J2 include all pixels from I1 and I2
% 'valid' - J1 and J2 are cropped to the size of the
% largest common rectangle containing valid pixels
%
% Default: 'valid'
%
% 'FillValues' An array containing one or several fill values.
% Fill values are used for output pixels when the
% corresponding inverse transformed location in the
% input image is completely outside the input image
% boundaries.
%
% If I1 and I2 are 2-D grayscale images then 'FillValues'
% must be a scalar. If I1 and I2 are a truecolor images,
% then 'FillValues' can be a scalar or a 3-element vector
% of RGB values.
%
% Default: 0
%
% Class Support
% -------------
% The class of I1 and I2 can be uint8, uint16, int16, double, single.
% I1 and I2 must be the of the same class. J1 and J2 are the same class
% as I1 and I2. stereoParams must be a stereoParameters object. tform1
% and tform2 must be 3-by-3 single or double projective transform matrices
% or projective2d objects.
%
% Example
% -------
% % Specify images containing a checkerboard for calibration
% imageDir = fullfile(toolboxdir('vision'), 'visiondata', ...
% 'calibration', 'stereo');
% leftImages = imageDatastore(fullfile(imageDir, 'left'));
% rightImages = imageDatastore(fullfile(imageDir, 'right'));
%
% % Detect the checkerboards
% [imagePoints, boardSize] = detectCheckerboardPoints(...
% leftImages.Files, rightImages.Files);
%
% % Specify world coordinates of checkerboard keypoints
% squareSizeInMillimeters = 108;
% worldPoints = generateCheckerboardPoints(boardSize, squareSizeInMillimeters);
%
% % Calibrate the stereo camera system
% stereoParams = estimateCameraParameters(imagePoints, worldPoints);
%
% % Read in the images
% I1 = readimage(leftImages, 1);
% I2 = readimage(rightImages, 1);
%
% % Rectify the images using 'full' output view
% [J1_full, J2_full] = rectifyStereoImages(I1, I2, stereoParams, ...
% 'OutputView', 'full');
%
% % Display the result
% figure;
% imshow(stereoAnaglyph(J1_full, J2_full));
%
% % Rectify the images using 'valid' output view. This is most suitable
% % for computing disparity.
% [J1_valid, J2_valid] = rectifyStereoImages(I1, I2, stereoParams, ...
% 'OutputView', 'valid');
%
% % Display the result
% figure;
% imshow(stereoAnaglyph(J1_valid, J2_valid));
%
% See also estimateCameraParameters, estimateUncalibratedRectification,
% disparity, reconstructScene, stereoParameters, cameraCalibrator,
% stereoCameraCalibrator
% Copyright 2014 MathWorks, Inc.
% References:
%
% G. Bradski and A. Kaehler, "Learning OpenCV : Computer Vision with
% the OpenCV Library," O'Reilly, Sebastopol, CA, 2008.
%#codegen
function [rectifiedImage1, rectifiedImage2] = rectifyStereoImages(I1, ...
I2, stereoParams, varargin)
vision.internal.inputValidation.validateImagePair(I1, I2, 'I1', 'I2');
if ~isa(stereoParams, 'stereoParameters')
narginchk(4, inf);
t1 = stereoParams;
t2 = varargin{1};
[tform1, tform2, interp, outputView, fillValues] = ...
parseInputsUncalibrated(I1, t1, t2, varargin{2:end});
[rectifiedImage1, rectifiedImage2] = rectifyUncalibrated(I1, I2, ...
tform1, tform2, interp, fillValues, outputView);
else
[interp, outputView, fillValues] = parseInputsCalibrated(...
I1, stereoParams, varargin{:});
[rectifiedImage1, rectifiedImage2] = rectifyStereoImagesImpl(stereoParams, ...
I1, I2, interp, fillValues, outputView);
end
%--------------------------------------------------------------------------
function [interp, outputView, fillValues] = ...
parseInputsCalibrated(image1, stereoParams, varargin)
validateStereoParameters(stereoParams);
if isempty(coder.target)
[interp, outputView, fillValues] = ...
vision.internal.inputValidation.parseUndistortRectifyInputsMatlab(...
'rectifyStereoImages', image1, @validateOutputViewPartial, varargin{:});
else
[interp, outputView, fillValues] = ...
vision.internal.inputValidation.parseUndistortRectifyInputsCodegen(...
image1, 'rectifyStereoImages', 'valid', varargin{:});
end
fillValues = vision.internal.inputValidation.scalarExpandFillValues(...
fillValues, image1);
%--------------------------------------------------------------------------
function [tform1, tform2, interp, outputView, fillValues] = ...
parseInputsUncalibrated(I1, t1, t2, varargin)
validateTransform(t1, 'tform1');
validateTransform(t2, 'tform2');
if isnumeric(t1)
tform1 = projective2d(t1);
tform2 = projective2d(t2);
else
tform1 = t1;
tform2 = t2;
end
if isempty(coder.target)
[interp, outputView, fillValues] = ...
vision.internal.inputValidation.parseUndistortRectifyInputsMatlab(...
'rectifyStereoImages', I1, @validateOutputViewPartial, varargin{:});
else
[interp, outputView, fillValues] = ...
vision.internal.inputValidation.parseUndistortRectifyInputsCodegen(...
I1, 'rectifyStereoImages', 'valid', varargin{:});
end
%--------------------------------------------------------------------------
function TF = validateTransform(tform, varName)
validateattributes(tform, {'double', 'single', 'projective2d'}, {}, ...
mfilename, varName); %#ok<EMCA>
if ~isa(tform, 'projective2d')
validateattributes(tform, {'double', 'single'}, ...
{'real', 'nonsparse', 'size', [3, 3]}, mfilename, varName); %#ok<EMCA>
end
TF = true;
%--------------------------------------------------------------------------
function TF = validateStereoParameters(params)
validateattributes(params, {'stereoParameters'}, ...
{'scalar'}, mfilename, 'stereoParams'); %#ok<EMCA>
TF = true;
%--------------------------------------------------------------------------
function outputView = validateOutputViewPartial(outputView)
outputView = ...
validatestring(outputView, {'full', 'valid'}, mfilename, 'OutputView'); %#ok<EMCA>
%--------------------------------------------------------------------------
function [J1, J2] = rectifyUncalibrated(I1, I2, tform1, tform2, interp, ...
fillValues, outputView)
% Compute the transformed location of image corners.
outPts = zeros(8, 2);
numRows = size(I1, 1);
numCols = size(I1, 2);
inPts = [1, 1; 1, numRows; numCols, numRows; numCols, 1];
outPts(1:4,1:2) = transformPointsForward(tform1, inPts);
numRows = size(I2, 1);
numCols = size(I2, 2);
inPts = [1, 1; 1, numRows; numCols, numRows; numCols, 1];
outPts(5:8,1:2) = transformPointsForward(tform2, inPts);
xSort = sort(outPts(:,1));
ySort = sort(outPts(:,2));
xLim = zeros(1, 2);
yLim = zeros(1, 2);
if strcmp(outputView, 'valid')
% Compute the common rectangular area of the transformed images.
xLim(1) = ceil(xSort(4)) - 0.5;
xLim(2) = floor(xSort(5)) + 0.5;
yLim(1) = ceil(ySort(4)) - 0.5;
yLim(2) = floor(ySort(5)) + 0.5;
else % full
xLim(1) = ceil(xSort(1)) - 0.5;
xLim(2) = floor(xSort(8)) + 0.5;
yLim(1) = ceil(ySort(1)) - 0.5;
yLim(2) = floor(ySort(8)) + 0.5;
end
width = xLim(2) - xLim(1) - 1;
height = yLim(2) - yLim(1) - 1;
outputViewRef = imref2d([height, width], xLim, yLim);
% Transform the images.
J1 = imwarp(I1, tform1, interp, 'OutputView', outputViewRef, 'FillValues', ...
fillValues);
J2 = imwarp(I2, tform2, interp, 'OutputView', outputViewRef, 'FillValues', ...
fillValues);