% * This code was used in the following articles: % * [1] Learning 3-D Scene Structure from a Single Still Image, % * Ashutosh Saxena, Min Sun, Andrew Y. Ng, % * In ICCV workshop on 3D Representation for Recognition (3dRR-07), 2007. % * (best paper) % * [2] 3-D Reconstruction from Sparse Views using Monocular Vision, % * Ashutosh Saxena, Min Sun, Andrew Y. Ng, % * In ICCV workshop on Virtual Representations and Modeling % * of Large-scale environments (VRML), 2007. % * [3] 3-D Depth Reconstruction from a Single Still Image, % * Ashutosh Saxena, Sung H. Chung, Andrew Y. Ng. % * International Journal of Computer Vision (IJCV), Aug 2007. % * [6] Learning Depth from Single Monocular Images, % * Ashutosh Saxena, Sung H. Chung, Andrew Y. Ng. % * In Neural Information Processing Systems (NIPS) 18, 2005. % * % * These articles are available at: % * http://make3d.stanford.edu/publications % * % * We request that you cite the papers [1], [3] and [6] in any of % * your reports that uses this code. % * Further, if you use the code in image3dstiching/ (multiple image version), % * then please cite [2]. % * % * If you use the code in third_party/, then PLEASE CITE and follow the % * LICENSE OF THE CORRESPONDING THIRD PARTY CODE. % * % * Finally, this code is for non-commercial use only. For further % * information and to obtain a copy of the license, see % * % * http://make3d.stanford.edu/publications/code % * % * Also, the software distributed under the License is distributed on an % * "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either % * express or implied. See the License for the specific language governing % * permissions and limitations under the License. % * % */ function [ Rc1_2, Rc2_1, ConS1_2, ConS2_1, RoughConS1_2, RoughConS2_1]=EffCalMatchSearchRegin(defaultPara, ScaleImg1, ScaleImg2, x1, x2, R, T, D1, D2, FlagDist); % This function Calculate the Constrain of SurfMatch search space % with different information given: % 1) Estimated Rotation and Translation matrix and depths % 2) Estimated Rotation and Translation matrix (no depths) % Return: % Rc1_2 - (4 x length(x1) : the 4 element of each column is the vectorized rotation matrix) % constrain for Img1 as Target, Img2 as Field % Rc2_1 - (4 x length(x2) : the 4 element of each column is the vectorized rotation matrix) % constrain for Img2 as Target, Img1 as Field % ConS1_2 - (4 x length(x1)) constrain for Img1 as Target, Img2 as Field % ConS1_2([1 2],:) - reference corner for the constrain square (x y) % ConS1_2([3],:) - sqare width along the epipolar line % ConS1_2([4],:) - sqare height othorgonal to the epipolar line % ConS2_1 - (4 x length(x2)) constrain for Img2 as Target, Img1 as Field % ConS2_1([1 2],:) - reference corner for the constrain square (x y) % ConS2_1([3],:) - sqare width along the epipolar line % ConS2_1([4],:) - sqare height othorgonal to the epipolar line % RoughConS1_2 - (4 x length(x1)) constrain for Img1 as Target, Img2 as Field (not allow rotation) % RoughConS1_2([1 2 3 4],:) - [xmax; xmin; ymax; ymin]; % RoughConS2_1 - (4 x length(x2)) constrain for Img2 as Target, Img1 as Field (not allow rotation) % RoughConS2_1([1 2 3 4],:) - [xmax; xmin; ymax; ymin]; % initialize Parameter NegativeDepthTolerence = defaultPara.NegativeDepthTolerence; MaxRatio =defaultPara.MaxRatio;%300; MinRatio = 1/MaxRatio; HeightImg1 = defaultPara.VertVar*max( ScaleImg1); HeightImg2 = defaultPara.VertVar*max( ScaleImg2); K1 = size(x1,2); K2 = size(x2,2); x1 = [x1; ones(1,K1)]; x2 = [x2; ones(1,K2)]; x1_calib = inv(defaultPara.InrinsicK1)*x1; x2_calib = inv(defaultPara.InrinsicK2)*x2; if isempty( R) || isempty( T) Rc1_2 = zeros(4, K1); Rc1_2( [1 4],:) = 1; Rc2_1 = zeros(4, K2); Rc2_1( [1 4],:) = 1; ConS1_2(1,:) = zeros(1, K1); ConS1_2(2,:) = ScaleImg2(2)/2*ones(1, K1); ConS1_2(3,:) = ScaleImg2(1); ConS1_2(4,:) = ScaleImg2(2)/2; ConS2_1(1,:) = zeros(1, K2); ConS2_1(2,:) = ScaleImg1(2)/2*ones(1, K2); ConS2_1(3,:) = ScaleImg1(1); ConS2_1(4,:) = ScaleImg1(2)/2; RoughConS1_2(1,:) = ScaleImg2(1)*ones(1,K1); RoughConS1_2(2,:) = 0; RoughConS1_2(3,:) = ScaleImg2(2)*ones(1,K1); RoughConS1_2(4,:) = 0; RoughConS2_1(1,:) = ScaleImg1(1)*ones(1,K2); RoughConS2_1(2,:) = 0; RoughConS2_1(3,:) = ScaleImg1(2)*ones(1,K2); RoughConS2_1(4,:) = 0; else R1_2 = R(1:3,:); R2_1 = R(4:6,:); T1_2 = T(1:3,:); T2_1 = T(4:6,:); T1_2_hat = [[0 -T1_2(3) T1_2(2)];... [T1_2(3) 0 -T1_2(1)];... [-T1_2(2) T1_2(1) 0]]; E = T1_2_hat*R1_2; F = inv(defaultPara.InrinsicK2')*E*inv(defaultPara.InrinsicK1) % I1 project on I2 ========================================== % 1) calculated the closed Depth and the Farest Depth that can be seen from Img2 % find Two End points of Epipolar line on Img2 [ EndPointsImg2 ] = EndPointsFromF(F, x1, ScaleImg2); [ EndPointsDepthImg1(1,:) dump Error] = triangulation( defaultPara, R1_2, T1_2, [x1; inv(defaultPara.InrinsicK1)*[EndPointsImg2(1:2,:); ones(1,K1)]]); [ EndPointsDepthImg1(2,:) dump Error] = triangulation( defaultPara, R1_2, T1_2, [x1; inv(defaultPara.InrinsicK2)*[EndPointsImg2(3:4,:); ones(1,K1)]]); EndPointsDepthImg1 = sort(EndPointsDepthImg1,1); % make the EndPointsDepthImg1 in acend order from top to bottom % 2) prune depth range if ~isempty( D1 ) MaxD1 = D1*MaxRatio; MinD1 = D1*MinRatio; % prune by EndPointsDepth MaxD1 = min(MaxD1, EndPointsDepthImg1(2,:)); MaxD1 = max(MaxD1, EndPointsDepthImg1(1,:)); MinD1 = max(MinD1, EndPointsDepthImg1(1,:)); MinD1 = min(MinD1, EndPointsDepthImg1(2,:)); else MaxD1 = EndPointsDepthImg1(2,:); MinD1 = EndPointsDepthImg1(1,:); end % prune by additional constrain ========OPtional MaxD1 = min(MaxD1, defaultPara.FarestDist); MinD1 = max(MinD1, defaultPara.Closestdist); % ============================================== % calculate the projection position x1CaMax3D = inv(defaultPara.InrinsicK1)*(x1.*repmat(MaxD1,3,1)); % 3-D position in camera 1 coordinate (3 by n) x1CaMin3D = inv(defaultPara.InrinsicK1)*(x1.*repmat(MinD1,3,1)); % 3-D position in camera 1 coordinate (3 by n) x1CaMaxHomo = [ x1CaMax3D; ones(1,K1)]; % into homogenous coordinate (4 by n) x1CaMinHomo = [ x1CaMin3D; ones(1,K1)]; % into homogenous coordinate (4 by n) x1_2Max3D = [R1_2 T1_2]*x1CaMaxHomo; % 3-D position in camera 2 coordinate (3 by n) x1_2MaxHomo = defaultPara.InrinsicK2*x1_2Max3D; % image homo coordinate in camera2 (3 by n) x1_2Max = [ x1_2MaxHomo(1,:)./x1_2MaxHomo(3,:); x1_2MaxHomo(2,:)./x1_2MaxHomo(3,:)]; % image coordinate (2 by n) x1_2Min3D = [R1_2 T1_2]*x1CaMinHomo; % 3-D position in camera 2 coordinate (3 by n) x1_2MinHomo = defaultPara.InrinsicK2*x1_2Min3D; % image homo coordinate in camera2 (3 by n) x1_2Min = [ x1_2MinHomo(1,:)./x1_2MinHomo(3,:); x1_2MinHomo(2,:)./x1_2MinHomo(3,:)]; % image coordinate (2 by n) % expand the search space a little bit in case the R and T are not accurate enough x1_2Max = x1_2Max + (x1_2Max - x1_2Min)*NegativeDepthTolerence;%Min529 x1_2Min = x1_2Min + (x1_2Min - x1_2Max)*NegativeDepthTolerence;%Min529 % Define Constrain (simple rectangle) [ Rc1_2, ConS1_2, RoughConS1_2 ]=Points2SqareConstrain( [ x1_2Max; x1_2Min], HeightImg1); % =========================================================== % I2 project on I1 ========================================== % 1) calculated the closed Depth and the Farest Depth that can be seen from Img1 % find Two End points of Epipolar line on Img2 [ EndPointsImg1 ] = EndPointsFromF(F', x2, ScaleImg1); [ EndPointsDepthImg2(1,:) dump Error] = triangulation( defaultPara, R2_1, T2_1, [x2; inv(defaultPara.InrinsicK1)*[EndPointsImg1(1:2,:); ones(1,K2)]]); [ EndPointsDepthImg2(2,:) dump Error] = triangulation( defaultPara, R2_1, T2_1, [x2; inv(defaultPara.InrinsicK2)*[EndPointsImg1(3:4,:); ones(1,K2)]]); EndPointsDepthImg2 = sort(EndPointsDepthImg2,1); % make the EndPointsDepthImg1 in acend order from top to bottom % 2) prune depth range if ~isempty( D2) MaxD2 = D2*MaxRatio; MinD2 = D2*MinRatio; % prune by EndPointsDepth MaxD2 = min(MaxD2, EndPointsDepthImg2(2,:)); MaxD2 = max(MaxD2, EndPointsDepthImg2(1,:)); MinD2 = max(MinD2, EndPointsDepthImg2(1,:)); MinD2 = min(MinD2, EndPointsDepthImg2(2,:)); else MaxD2 = EndPointsDepthImg2(2,:); MinD2 = EndPointsDepthImg2(1,:); end % prune by additional constrain ========OPtional MaxD2 = min(MaxD2, defaultPara.FarestDist); MinD2 = max(MinD2, defaultPara.Closestdist); % ============================================== % calculate the projection position x2CaMax3D = inv(defaultPara.InrinsicK2)*(x2.*repmat(MaxD2,3,1)); % 3-D position in camera 1 coordinate (3 by n) x2CaMin3D = inv(defaultPara.InrinsicK2)*(x2.*repmat(MinD2,3,1)); % 3-D position in camera 1 coordinate (3 by n) x2CaMaxHomo = [ x2CaMax3D; ones(1,K2)]; % into homogenous coordinate (4 by n) x2CaMinHomo = [ x2CaMin3D; ones(1,K2)]; % into homogenous coordinate (4 by n) x2_1Max3D = [R2_1 T2_1]*x2CaMaxHomo; % 3-D position in camera 2 coordinate (3 by n) x2_1MaxHomo = defaultPara.InrinsicK2*x2_1Max3D; % image homo coordinate in camera2 (3 by n) x2_1Max = [ x2_1MaxHomo(1,:)./x2_1MaxHomo(3,:); x2_1MaxHomo(2,:)./x2_1MaxHomo(3,:)]; % image coordinate (2 by n) x2_1Min3D = [R2_1 T2_1]*x2CaMinHomo; % 3-D position in camera 2 coordinate (3 by n) x2_1MinHomo = defaultPara.InrinsicK2*x2_1Min3D; % image homo coordinate in camera2 (3 by n) x2_1Min = [ x2_1MinHomo(1,:)./x2_1MinHomo(3,:); x2_1MinHomo(2,:)./x2_1MinHomo(3,:)]; % image coordinate (2 by n) % expand the search space a little bit in case the R and T are not accurate enough x2_1Max = x2_1Max + (x2_1Max - x2_1Min)*NegativeDepthTolerence;%Min529 x2_1Min = x2_1Min + (x2_1Min - x2_1Max)*NegativeDepthTolerence;%Min529 % Define Constrain (simple rectangle) [ Rc2_1, ConS2_1, RoughConS2_1 ]=Points2SqareConstrain( [ x2_1Max; x2_1Min], HeightImg2); % =========================================================== if FlagDisp %figure; %dispMatchSearchRegin(I1, I2, x1, x2, tempConS1_2, tempConS2_1, F, ... %x1_2Max, MaxD1, x1_2Min, MinD1, ... %x2_1Max, MaxD2, x2_1Min, MinD2, ... %FlagRotate, 'Stacking', 'h', 'Interactive', 0); figure; dispMatchSearchRegin(I1, I2, x1, x2, tempConSConS1_2, tempConSConS2_1, F, FlagRotate, 'Stacking', 'v', 'Interactive', 0); end end end % return