Correct spelling in comments and docs.

Change-Id: Iad9a0599d644d3b3cd54244edaf64d408cb1308e

examples/bal_problem.cc

@@ -78,7 +78,7 @@ BALProblem::BALProblem(const std::string& filename, bool use_quaternions) {
     return;
   };
 
-  // This wil die horribly on invalid files. Them's the breaks.
+  // This will die horribly on invalid files. Them's the breaks.
   FscanfOrDie(fptr, "%d", &num_cameras_);
   FscanfOrDie(fptr, "%d", &num_points_);
   FscanfOrDie(fptr, "%d", &num_observations_);

examples/bundle_adjuster.cc

@@ -120,7 +120,7 @@ DEFINE_double(point_sigma, 0.0, "Standard deviation of the point "
               "perturbation.");
 DEFINE_int32(random_seed, 38401, "Random seed used to set the state "
              "of the pseudo random number generator used to generate "
-             "the pertubations.");
+             "the perturbations.");
 DEFINE_bool(line_search, false, "Use a line search instead of trust region "
             "algorithm.");
 DEFINE_bool(mixed_precision_solves, false, "Use mixed precision solves.");
@@ -298,7 +298,7 @@ void BuildProblem(BALProblem* bal_problem, Problem* problem) {
     LossFunction* loss_function =
         CERES_GET_FLAG(FLAGS_robustify) ? new HuberLoss(1.0) : nullptr;
 
-    // Each observation correponds to a pair of a camera and a point
+    // Each observation corresponds to a pair of a camera and a point
     // which are identified by camera_index()[i] and point_index()[i]
     // respectively.
     double* camera =

examples/denoising.cc

@@ -33,7 +33,7 @@
 // Note that for good denoising results the weighting between the data term
 // and the Fields of Experts term needs to be adjusted. This is discussed
 // in [1]. This program assumes Gaussian noise. The noise model can be changed
-// by substituing another function for QuadraticCostFunction.
+// by substituting another function for QuadraticCostFunction.
 //
 // [1] S. Roth and M.J. Black. "Fields of Experts." International Journal of
 //     Computer Vision, 82(2):205--229, 2009.

examples/fields_of_experts.cc

@@ -28,7 +28,7 @@
 //
 // Author: strandmark@google.com (Petter Strandmark)
 //
-// Class for loading the data required for descibing a Fields of Experts (FoE)
+// Class for loading the data required for describing a Fields of Experts (FoE)
 // model.
 
 #include "fields_of_experts.h"

examples/fields_of_experts.h

@@ -28,7 +28,7 @@
 //
 // Author: strandmark@google.com (Petter Strandmark)
 //
-// Class for loading the data required for descibing a Fields of Experts (FoE)
+// Class for loading the data required for describing a Fields of Experts (FoE)
 // model. The Fields of Experts regularization consists of terms of the type
 //
 //   alpha * log(1 + (1/2)*sum(F .* X)^2),

examples/libmv_bundle_adjuster.cc

@@ -60,7 +60,7 @@
 // Image number shall be greater or equal to zero. Order of cameras does not
 // matter and gaps are possible.
 //
-// Every 3D point is decribed by:
+// Every 3D point is described by:
 //
 //  - Track number point belongs to (single 4 bytes integer value).
 //  - 3D position vector, 3-component vector of float values.

examples/libmv_homography.cc

@@ -380,10 +380,10 @@ int main(int argc, char** argv) {
 
   Mat x2 = x1;
   for (int i = 0; i < x2.cols(); ++i) {
-    Vec3 homogenous_x1 = Vec3(x1(0, i), x1(1, i), 1.0);
-    Vec3 homogenous_x2 = homography_matrix * homogenous_x1;
-    x2(0, i) = homogenous_x2(0) / homogenous_x2(2);
-    x2(1, i) = homogenous_x2(1) / homogenous_x2(2);
+    Vec3 homogeneous_x1 = Vec3(x1(0, i), x1(1, i), 1.0);
+    Vec3 homogeneous_x2 = homography_matrix * homogeneous_x1;
+    x2(0, i) = homogeneous_x2(0) / homogeneous_x2(2);
+    x2(1, i) = homogeneous_x2(1) / homogeneous_x2(2);
 
     // Apply some noise so algebraic estimation is not good enough.
     x2(0, i) += static_cast<double>(rand() % 1000) / 5000.0;

examples/nist.cc

@@ -117,7 +117,7 @@ DEFINE_int32(max_line_search_restarts,
              "Maximum number of restarts of line search direction algorithm.");
 DEFINE_string(line_search_interpolation,
               "cubic",
-              "Degree of polynomial aproximation in line search, choices are: "
+              "Degree of polynomial approximation in line search, choices are: "
               "bisection, quadratic & cubic.");
 DEFINE_int32(lbfgs_rank,
              20,
@@ -179,7 +179,7 @@ void SplitStringUsingChar(const string& full,
     } else {
       const char* start = p;
       while (++p != end && *p != delim) {
-        // Skip to the next occurence of the delimiter.
+        // Skip to the next occurrence of the delimiter.
       }
       *it++ = string(start, p - start);
     }
@@ -250,7 +250,7 @@ class NISTProblem {
       final_parameters_(0, parameter_id) = atof(pieces[2 + kNumTries].c_str());
     }
 
-    // Certfied cost
+    // Certified cost
     SkipLines(ifs, 1);
     GetAndSplitLine(ifs, &pieces);
     certified_cost_ = atof(pieces[4].c_str()) / 2.0;

examples/robot_pose_mle.cc

@@ -50,7 +50,7 @@
 //
 // There are two types of residuals in this problem:
 // 1) The OdometryConstraint residual, that accounts for the odometry readings
-//    between successive pose estimatess of the robot.
+//    between successive pose estimates of the robot.
 // 2) The RangeConstraint residual, that accounts for the errors in the observed
 //    range readings from each pose.
 //
@@ -97,14 +97,14 @@
 // timesteps 0 to i for that variable, both inclusive.
 //
 // Bayes' rule is used to derive eq. 3 from 2, and the independence of
-// odometry observations and range readings is expolited to derive 4 from 3.
+// odometry observations and range readings is exploited to derive 4 from 3.
 //
 // Thus, the Belief, up to scale, is factored as a product of a number of
 // terms, two for each pose, where for each pose term there is one term for the
 // range reading, P(y_i | u*_(0:i) and one term for the odometry reading,
 // P(u*_i | u_i) . Note that the term for the range reading is dependent on all
 // odometry values u*_(0:i), while the odometry term, P(u*_i | u_i) depends only
-// on a single value, u_i. Both the range reading as well as odoemtry
+// on a single value, u_i. Both the range reading as well as odometry
 // probability terms are modeled as the Normal distribution, and have the form:
 //
 // p(x) \propto \exp{-((x - x_mean) / x_stddev)^2}
@@ -123,7 +123,7 @@
 // variable, and will be computed by an AutoDiffCostFunction, while the term
 // for the range reading will depend on all previous odometry observations, and
 // will be computed by a DynamicAutoDiffCostFunction since the number of
-// odoemtry observations will only be known at run time.
+// odometry observations will only be known at run time.
 
 #include <cmath>
 #include <cstdio>

examples/slam/common/read_g2o.h

@@ -59,7 +59,7 @@ bool ReadVertex(std::ifstream* infile,
   return true;
 }
 
-// Reads the contraints between two vertices in the pose graph
+// Reads the constraints between two vertices in the pose graph
 template <typename Constraint, typename Allocator>
 void ReadConstraint(std::ifstream* infile,
                     std::vector<Constraint, Allocator>* constraints) {