svo
Semi-Direct Visual Odometry
|
svo::feature_detection::AbstractDetector | All detectors should derive from this abstract class |
svo::Reprojector::Candidate | A candidate is a point that projects into the image plane and for which we will search a maching feature in the image |
svo::Config | Global configuration file of SVO |
svo::feature_detection::Corner | Temporary container used for corner detection. Features are initialized from these |
svo::DepthFilter | Depth filter implements the Bayesian Update proposed in: "Video-based, Real-Time Multi View Stereo" by G |
svo::ba::EdgeContainerSE3 | Temporary container to hold the g2o edge with reference to frame and point |
svo::feature_detection::FastDetector | FAST detector by Edward Rosten |
svo::Feature | A salient image region that is tracked across frames |
svo::Frame | A frame saves the image, the associated features and the estimated pose |
svo::FrameHandlerBase | Base class for various VO pipelines. Manages the map and the state machine |
svo::FrameHandlerMono | Monocular Visual Odometry Pipeline as described in the SVO paper |
svo::Reprojector::Grid | The grid stores a set of candidate matches. For every grid cell we try to find one match |
svo::initialization::KltHomographyInit | Tracks features using Lucas-Kanade tracker and then estimates a homography |
svo::Map | Map object which saves all keyframes which are in a map |
svo::MapPointCandidates | Container for converged 3D points that are not already assigned to two keyframes |
svo::Matcher | Patch-matcher for reprojection-matching and epipolar search in triangulation |
svo::Matcher::Options | |
svo::DepthFilter::Options | Depth-filter config parameters |
svo::Reprojector::Options | Reprojector config parameters |
svo::Point | A 3D point on the surface of the scene |
svo::Reprojector | Project points from the map into the image and find the corresponding feature (corner) |
svo::Seed | A seed is a probabilistic depth estimate for a single pixel |
svo::SparseImgAlign | Optimize the pose of the frame by minimizing the photometric error of feature patches |