[37] | 1 | % * This code was used in the following articles:
|
---|
| 2 | % * [1] Learning 3-D Scene Structure from a Single Still Image,
|
---|
| 3 | % * Ashutosh Saxena, Min Sun, Andrew Y. Ng,
|
---|
| 4 | % * In ICCV workshop on 3D Representation for Recognition (3dRR-07), 2007.
|
---|
| 5 | % * (best paper)
|
---|
| 6 | % * [2] 3-D Reconstruction from Sparse Views using Monocular Vision,
|
---|
| 7 | % * Ashutosh Saxena, Min Sun, Andrew Y. Ng,
|
---|
| 8 | % * In ICCV workshop on Virtual Representations and Modeling
|
---|
| 9 | % * of Large-scale environments (VRML), 2007.
|
---|
| 10 | % * [3] 3-D Depth Reconstruction from a Single Still Image,
|
---|
| 11 | % * Ashutosh Saxena, Sung H. Chung, Andrew Y. Ng.
|
---|
| 12 | % * International Journal of Computer Vision (IJCV), Aug 2007.
|
---|
| 13 | % * [6] Learning Depth from Single Monocular Images,
|
---|
| 14 | % * Ashutosh Saxena, Sung H. Chung, Andrew Y. Ng.
|
---|
| 15 | % * In Neural Information Processing Systems (NIPS) 18, 2005.
|
---|
| 16 | % *
|
---|
| 17 | % * These articles are available at:
|
---|
| 18 | % * http://make3d.stanford.edu/publications
|
---|
| 19 | % *
|
---|
| 20 | % * We request that you cite the papers [1], [3] and [6] in any of
|
---|
| 21 | % * your reports that uses this code.
|
---|
| 22 | % * Further, if you use the code in image3dstiching/ (multiple image version),
|
---|
| 23 | % * then please cite [2].
|
---|
| 24 | % *
|
---|
| 25 | % * If you use the code in third_party/, then PLEASE CITE and follow the
|
---|
| 26 | % * LICENSE OF THE CORRESPONDING THIRD PARTY CODE.
|
---|
| 27 | % *
|
---|
| 28 | % * Finally, this code is for non-commercial use only. For further
|
---|
| 29 | % * information and to obtain a copy of the license, see
|
---|
| 30 | % *
|
---|
| 31 | % * http://make3d.stanford.edu/publications/code
|
---|
| 32 | % *
|
---|
| 33 | % * Also, the software distributed under the License is distributed on an
|
---|
| 34 | % * "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either
|
---|
| 35 | % * express or implied. See the License for the specific language governing
|
---|
| 36 | % * permissions and limitations under the License.
|
---|
| 37 | % *
|
---|
| 38 | % */
|
---|
| 39 | function [matches1 matches2] = GenMatches(defaultPara, ImgInfo, FlagDisp) |
---|
| 40 | |
---|
| 41 | % This function generate Matches uing IMU and GPS info and Ransac and BA |
---|
| 42 | |
---|
| 43 | % 1. Mono calulation or load the pre-calculated data ------------------------ |
---|
| 44 | ImgInfo(1).appendOpt = 0; |
---|
| 45 | ImgInfo(2).appendOpt = 0; |
---|
| 46 | [ ImgInfo] = SingleModelInfo(defaultPara, ImgInfo); |
---|
| 47 | |
---|
| 48 | % initialize variables |
---|
| 49 | Img1 = strrep(ImgInfo(1).ExifInfo.name,'.jpg',''); |
---|
| 50 | Img2 = strrep(ImgInfo(2).ExifInfo.name,'.jpg',''); |
---|
| 51 | I1=imreadbw([defaultPara.Fdir '/pgm/' Img1 '.pgm']); % function from sift |
---|
| 52 | I2=imreadbw([defaultPara.Fdir '/pgm/' Img2 '.pgm']); % function from sift |
---|
| 53 | [f1] = readSurf(Img1, defaultPara.Fdir, 'Dense'); % original features |
---|
| 54 | [f2] = readSurf(Img2, defaultPara.Fdir, 'Dense'); % original features |
---|
| 55 | [D1 IND] = PorjPosi2Depth(size(I1), size(ImgInfo(1).Model.Depth.FitDepth), f1, ImgInfo(1).Model.Depth.FitDepth); |
---|
| 56 | [D2 IND] = PorjPosi2Depth(size(I2), size(ImgInfo(2).Model.Depth.FitDepth), f2, ImgInfo(1).Model.Depth.FitDepth); |
---|
| 57 | |
---|
| 58 | % 1. extract Measuesd Position and orientation from GPS or IMU info |
---|
| 59 | [MeasR MeasT] = InitPoseMeas(defaultPara, ImgInfo(1), ImgInfo(2)); |
---|
| 60 | |
---|
| 61 | % 2. Using Measures R and T and Mono-Depth to define mach search space constrain |
---|
| 62 | % read in all surf features |
---|
| 63 | [ Rc1, Rc2, ConS1, ConS2, ConSRough1, ConSRough2] = CalMatchSearchRegin(defaultPara, MeasR, MeasT, I1, I2, f1, f2, D1, D2, 1, FlagDisp); |
---|
| 64 | Vector2Ipoint([Rc1; ConS1],[defaultPara.Fdir '/surf/'],['RConS_' Img1]); |
---|
| 65 | Vector2Ipoint([Rc2; ConS2],[defaultPara.Fdir '/surf/'],['RConS_' Img2]); |
---|
| 66 | Vector2Ipoint([ConSRough1],[defaultPara.Fdir '/surf/'],['RConSRough_' Img1]); |
---|
| 67 | Vector2Ipoint([ConSRough2],[defaultPara.Fdir '/surf/'],['RConSRough_' Img2]); |
---|
| 68 | |
---|
| 69 | % 3. Do match search with all combinations satisfying Constrain from 2) using ralative threshould |
---|
| 70 | tic; |
---|
| 71 | cd match |
---|
| 72 | pwd |
---|
| 73 | % system(['./surfMatchRConS.sh ' defaultPara.Fdir ' ' Img1 ' ' Img2 ' _ 0.3 0.7']); |
---|
| 74 | system(['./surfMatchRConS.sh ' defaultPara.Fdir ' ' Img1 ' ' Img2 ' Dense ' '0.3 0.7']); % Parameter still need to be changed//Min |
---|
| 75 | cd .. |
---|
| 76 | toc |
---|
| 77 | |
---|
| 78 | % 4. Ransac |
---|
| 79 | [f1, f2, matches] = readSurfMatches(Img1, Img2, defaultPara.Fdir, [ defaultPara.Type 'Dense'], 1, 1); |
---|
| 80 | if isempty(matches) |
---|
| 81 | disp('Zeros matches'); |
---|
| 82 | matches1 = matches(1,:); |
---|
| 83 | matches2 = matches(2,:); |
---|
| 84 | return; |
---|
| 85 | end |
---|
| 86 | [D1 IND1] = PorjPosi2Depth(size(I1), size(ImgInfo(1).Model.Depth.FitDepth), f1(:,matches(1,:)), ImgInfo(1).Model.Depth.FitDepth); |
---|
| 87 | [D2 IND2] = PorjPosi2Depth(size(I2), size(ImgInfo(2).Model.Depth.FitDepth), f2(:,matches(2,:)), ImgInfo(1).Model.Depth.FitDepth); |
---|
| 88 | %figure(11); plotmatches(I1,I2,f1, f2,matches, 'Stacking','v','Interactive', FlagDisp); title('SurfMatch') |
---|
| 89 | %saveas(11,[defaultPara.ScratchFolder Img1 '_' Img2 'SimpleSurfMatch'],'jpg'); |
---|
| 90 | [F, inliers, NewDist, fail]=GeneralRansac(defaultPara, f1, f2, matches, D1, D2); |
---|
| 91 | figure(12); plotmatches(I1,I2,f1, f2,matches(:,inliers), 'Stacking', 'v', 'Interactive', FlagDisp); |
---|
| 92 | saveas(12,[defaultPara.ScratchFolder Img1 '_' Img2 'AfterRansac'],'jpg'); |
---|
| 93 | close 12; |
---|
| 94 | |
---|
| 95 | % *** Stop maunally to pick out the bad matches*** ----------------- |
---|
| 96 | matches = matches(:,inliers); |
---|
| 97 | if isempty(matches) |
---|
| 98 | disp('Zeros matches'); |
---|
| 99 | matches1 = matches(1,:); |
---|
| 100 | matches2 = matches(2,:); |
---|
| 101 | return; |
---|
| 102 | end |
---|
| 103 | |
---|
| 104 | % x_calib = [ inv(defaultPara.InrinsicK1)*[ f1(:,matches(1,:)); ones(1,length(matches))];... |
---|
| 105 | % inv(defaultPara.InrinsicK2)*[ f2(:,matches(2,:)); ones(1,length(matches))]]; |
---|
| 106 | % [ lamda1 lamda2] = triangulation( defaultPara, MeasR(1:3,:), MeasT(1:3), x_calib); |
---|
| 107 | % % end |
---|
| 108 | % X_obj_1 = x_calib(1:3,:).*repmat(lamda1, 3, 1); |
---|
| 109 | % X_obj_2 = MeasR(4:6,:)*(x_calib(4:6,:).*repmat(lamda2, 3, 1)) + repmat(MeasT(4:6), 1, length(matches)); |
---|
| 110 | % X_obj = (X_obj_1+X_obj_2)/2; |
---|
| 111 | % %end |
---|
| 112 | |
---|
| 113 | % 5. Bundle Adjustment |
---|
| 114 | % [R T X_obj_BA X_im_BA dist1_BA dist2_BA]=SparseBAWraper(defaultPara, MeasR, MeasT, [f1(:,matches(1,:)); f2(:,matches(2,:))], X_obj, ImgInfo, 1); |
---|
| 115 | % outlier_thre1 = prctile(dist1_BA,90); |
---|
| 116 | % outlier_thre2 = prctile(dist2_BA,90); |
---|
| 117 | % Outlier = dist1_BA > outlier_thre1 | dist2_BA > outlier_thre2; |
---|
| 118 | % lamda1(Outlier) = []; |
---|
| 119 | % lamda2(Outlier) = []; |
---|
| 120 | % X_obj_BA(:,Outlier) = []; |
---|
| 121 | % x_calib(:,Outlier) = []; |
---|
| 122 | % matches(:, Outlier) = []; |
---|
| 123 | % % [R T X_obj_BA X_im_BA dist1_BA dist2_BA]=SparseBAWraper(defaultPara, R, T, [f1(:,matches(1,:)); f2(:,matches(2,:))], X_obj_BA, ImgInfo, 1); |
---|
| 124 | % figure(13); plotmatches(I1,I2,f1, f2,matches, 'Stacking', 'v', 'Interactive', FlagDisp);title('after BA clean once'); |
---|
| 125 | % saveas(13,[defaultPara.ScratchFolder Img1 '_' Img2 'AfterBA'],'jpg'); |
---|
| 126 | |
---|
| 127 | matches1 = matches(1,:); |
---|
| 128 | matches2 = matches(2,:); |
---|
| 129 | return; |
---|