Add various video reconstruction improvements.

More precisely, this commit concerns mainly the reconstruction lib. It
- adds 2 new functions in mapping for calibrated/uncalibrated traingulation
- improves the keyframe selection using a method based on the number of shared tracks
- adds two functions for processing keyframes and non keyframes in the euclidean reconstruction file
and also
- updates nview and some tools
- the 'end of lines' of several files are updated (to UNIX). 
- the warning occuring in the RGB image conversion to gray function is solved.
- updates License and Readme 
- removes CMakeLists.txt, we should use the one in src/.

LICENSE

@@ -1,4 +1,4 @@
-Copyright (c) 2007, 2008 libmv authors.
+Copyright (c) 2007-2011 libmv authors.
 
 Permission is hereby granted, free of charge, to any person obtaining a copy
 of this software and associated documentation files (the "Software"), to

README

@@ -52,9 +52,9 @@ Instead do this:
 
  $ ls
  AUTHORS  contrib  doc  LICENSE  Makefile  README  src
- $ mkdir alternate_build
- $ cd alternate_build
- $ cmake ../ # Your options here
+ $ mkdir build
+ $ cd build
+ $ cmake ../src # Your options here
  $ make
  $ make test
  $ ...

src/libmv/correspondence/ArrayMatcher_Kdtree_Flann.h

@@ -54,7 +54,7 @@ class ArrayMatcher_Kdtree_Flann : public ArrayMatcher<Scalar>
   bool build( const Scalar * dataset, int nbRows, int dimension)  {
 
     _p.log_destination = NULL;
-    _p.log_level = LOG_INFO;
+    _p.log_level = LOG_WARN;//LOG_INFO;
 
     // Force KDTREE matching
     _p.algorithm = KDTREE;

src/libmv/correspondence/robust_tracker.cc

@@ -142,8 +142,8 @@ bool RobustTracker::Track(const Image &image,
                                                  rms_threshold_inlier_, 
                                                  &F,
                                                  &inliers);
-      LOG(INFO) << "#inliers = "<< inliers.size() << std::endl;
-      LOG(INFO) << "#outliers = "<< tracks.size()-inliers.size() << std::endl;
+      VLOG(2) << "#inliers = "<< inliers.size() << std::endl;
+      VLOG(2) << "#outliers = "<< tracks.size()-inliers.size() << std::endl;
       // We remove correspondences that are not inliers
       size_t max_num_track = 1 +
        std::max(new_features_graph->matches_.GetMaxTrackID(),

src/libmv/image/image_converter.h

@@ -31,7 +31,7 @@ template<typename T>
 // var_R * 0.2126 + var_G * 0.7152 + var_B * 0.0722
 inline T RGB2GRAY(const T r,const T g, const T b)
 {
-  return r * 0.2126 + g * 0.7152 + b * 0.0722;
+  return static_cast<T>(r * 0.2126 + g * 0.7152 + b * 0.0722);
 }
 /*
 // Specialization for the uchar type. (that do not want to work...)

src/libmv/multiview/bundle.cc

@@ -49,7 +49,7 @@ showErrorStatistics(double const f0,
     Vector2d p = cams[i].projectPoint(distortion, Xs[j]);
     meanReprojectionError += Square(norm_L2(f0 * (p - measurements[k])));
   }
-  VLOG(2) << "mean reprojection error (in pixels): " 
+  VLOG(3) << "mean reprojection error (in pixels): " 
             << sqrt(meanReprojectionError/K) << std::endl;
   return sqrt(meanReprojectionError/K);
 }
@@ -108,7 +108,7 @@ void EuclideanBAFull(const vector<Mat2X> &x,
   distortion.p2 = 0;
 
   double const f0 = KMat[0][0];
-  VLOG(2) << "intrinsic before bundle = "; displayMatrix(KMat);
+  VLOG(3) << "intrinsic before bundle = "; displayMatrix(KMat);
   Matrix3x3d Knorm = KMat;
   // Normalize the intrinsic to have unit focal length.
   scaleMatrixIP(1.0/f0, Knorm);
@@ -170,7 +170,7 @@ void EuclideanBAFull(const vector<Mat2X> &x,
   double const inlierThreshold = 2.0 / f0;
 
   Matrix3x3d K0 = cams[0].getIntrinsic();
-  VLOG(2) << "K0 = "; displayMatrix(K0);
+  VLOG(3) << "K0 = "; displayMatrix(K0);
 
   CommonInternalsMetricBundleOptimizer opt(mode, inlierThreshold,
                                            K0, distortion,
@@ -179,17 +179,17 @@ void EuclideanBAFull(const vector<Mat2X> &x,
                                            correspondingPoint);
   opt.maxIterations = 50;
   opt.minimize();
-  VLOG(2) << "optimizer status = " << opt.status << endl;
-  VLOG(2) << "refined K = "; displayMatrix(K0);
-  VLOG(2) << "distortion = " << distortion.k1 << " " << distortion.k2 << " "
+  VLOG(3) << "optimizer status = " << opt.status << endl;
+  VLOG(3) << "refined K = "; displayMatrix(K0);
+  VLOG(3) << "distortion = " << distortion.k1 << " " << distortion.k2 << " "
             << distortion.p1 << " " << distortion.p2 << endl;
 
   for (int i = 0; i < num_camsN; ++i) cams[i].setIntrinsic(K0);
 
   Matrix3x3d Knew = K0;
   scaleMatrixIP(f0, Knew);
   Knew[2][2] = 1.0;
-  VLOG(2) << "Knew = "; displayMatrix(Knew);
+  VLOG(3) << "Knew = "; displayMatrix(Knew);
 
   showErrorStatistics(f0, distortion, cams, Xs, measurements, correspondingView,
                       correspondingPoint);
@@ -268,7 +268,7 @@ double EuclideanBA(const vector<Mat2X> &x,
   distortion.p2 = 0;
 
   double f0 = KMat[0][0];
-  VLOG(2) << "intrinsic before bundle = "; displayMatrix(KMat);
+  VLOG(3) << "intrinsic before bundle = "; displayMatrix(KMat);
   Matrix3x3d Knorm = KMat;
   // Normalize the intrinsic to have unit focal length.
   scaleMatrixIP(1.0/f0, Knorm);
@@ -344,7 +344,7 @@ double EuclideanBA(const vector<Mat2X> &x,
   double const inlierThreshold = 2.0 / f0;
 
   Matrix3x3d K0 = cams[0].getIntrinsic();
-  VLOG(2) << "K0 = "; displayMatrix(K0);
+  VLOG(3) << "K0 = "; displayMatrix(K0);
 
   CommonInternalsMetricBundleOptimizer opt(mode, inlierThreshold,
                                            K0, distortion,
@@ -353,17 +353,17 @@ double EuclideanBA(const vector<Mat2X> &x,
                                            correspondingPoint);
   opt.maxIterations = 50;
   opt.minimize();
-  VLOG(2) << "optimizer status = " << opt.status << std::endl;
-  VLOG(2) << "refined K = "; displayMatrix(K0);
-  VLOG(2) << "distortion = " << distortion.k1 << " " << distortion.k2 << " "
+  VLOG(3) << "optimizer status = " << opt.status << std::endl;
+  VLOG(3) << "refined K = "; displayMatrix(K0);
+  VLOG(3) << "distortion = " << distortion.k1 << " " << distortion.k2 << " "
             << distortion.p1 << " " << distortion.p2 << std::endl;
 
   for (int i = 0; i < num_camsN; ++i) cams[i].setIntrinsic(K0);
 
   Matrix3x3d Knew = K0;
   scaleMatrixIP(f0, Knew);
   Knew[2][2] = 1.0;
-  VLOG(2) << "Knew = "; displayMatrix(Knew);
+  VLOG(3) << "Knew = "; displayMatrix(Knew);
 
   double rms = showErrorStatistics(f0, distortion, cams, 
                                    Xs, measurements,correspondingView,

src/libmv/multiview/euclidean_resection.cc

@@ -442,14 +442,14 @@ void EuclideanResectionEPnP(const Mat2X &x_camera, const Mat3X &X_world,
     ui = x_camera(0, c);
     vi = x_camera(1, c);
     M.block(2*c, 0, 2, 12) << a0, 0, 
-                  a0*(-ui), a1, 0,
-                  a1*(-ui), a2, 0, 
-                  a2*(-ui), a3, 0,
-                  a3*(-ui), 0, 
-                  a0, a0*(-vi), 0,
-                  a1, a1*(-vi), 0,
-                  a2, a2*(-vi), 0,
-                  a3, a3*(-vi);
+                              a0*(-ui), a1, 0,
+                              a1*(-ui), a2, 0, 
+                              a2*(-ui), a3, 0,
+                              a3*(-ui), 0, 
+                              a0, a0*(-vi), 0,
+                              a1, a1*(-vi), 0,
+                              a2, a2*(-vi), 0,
+                              a3, a3*(-vi);
   }
 
   Eigen::SVD<Mat> MtMsvd = (M.transpose()*M).svd();

src/libmv/multiview/robust_estimation.h

@@ -112,13 +112,13 @@ typename Kernel::Model Estimate(const Kernel &kernel,
 
     vector<typename Kernel::Model> models;
     kernel.Fit(sample, &models);
-    VLOG(2) << "Fitted subset; found " << models.size() << " model(s).";
+    VLOG(4) << "Fitted subset; found " << models.size() << " model(s).";
 
     // Compute costs for each fit.
     for (int i = 0; i < models.size(); ++i) {
       vector<int> inliers;
       double cost = scorer.Score(kernel, models[i], all_samples, &inliers);
-      VLOG(3) << "Fit cost: " << cost
+      VLOG(5) << "Fit cost: " << cost
               << ", number of inliers: " << inliers.size();
 
       if (cost < best_cost) {
@@ -129,7 +129,7 @@ typename Kernel::Model Estimate(const Kernel &kernel,
         if (best_inliers) {
           best_inliers->swap(inliers);
         }
-        VLOG(2) << "New best cost: " << best_cost << " with "
+        VLOG(4) << "New best cost: " << best_cost << " with "
                 << best_num_inliers << " inlying of "
                 << total_samples << " total samples.";
       }
@@ -138,7 +138,7 @@ typename Kernel::Model Estimate(const Kernel &kernel,
                                             outliers_probability,
                                             best_inlier_ratio);
 
-      VLOG(2) << "Max iterations needed given best inlier ratio: "
+      VLOG(5) << "Max iterations needed given best inlier ratio: "
         << max_iterations << "; best inlier ratio: " << best_inlier_ratio;
     }
   }

src/libmv/reconstruction/CMakeLists.txt

@@ -2,19 +2,20 @@ ADD_LIBRARY(reconstruction
             euclidean_reconstruction.cc
             export_blender.cc
             export_ply.cc
-            image_order_selection.cc
+            image_selection.cc
+            keyframe_selection.cc
             mapping.cc
             optimization.cc
             projective_reconstruction.cc
             reconstruction.cc
             tools.cc
             )
 
-TARGET_LINK_LIBRARIES(reconstruction camera multiview numeric V3D colamd ldl)
+TARGET_LINK_LIBRARIES(reconstruction camera multiview numeric V3D colamd ldl glog)
 LIBMV_INSTALL_LIB(reconstruction)
 
 MACRO (RECONSTRUCTION_TEST NAME)
-  LIBMV_TEST(${NAME} "reconstruction;multiview_test_data;camera;correspondence;multiview;numeric")
+  LIBMV_TEST(${NAME} "reconstruction;multiview_test_data;camera;correspondence;multiview;numeric;glog")
 ENDMACRO (RECONSTRUCTION_TEST)
 
-RECONSTRUCTION_TEST(reconstruction)
+RECONSTRUCTION_TEST(euclidean_reconstruction)