Class Frame

Defined in File frame.hpp

Class Documentation

class Frame

Class for the camera frame.

Public Types

using shared_ptr = std::shared_ptr<Frame>: Shared pointer alias for Frame.

using weak_ptr = std::weak_ptr<Frame>: Weak pointer alias for Frame.

Public Functions

Frame(const data_type::SensorData::weak_ptr cameraData_)

Construct a frame and initialize image features from sensor input.

Parameters:: cameraData_ – Weak pointer to sensor data for this frame.

inline ~Frame(): Destroy the frame object.

inline bool is_monocular() const

Checks if the camera data is from a monocular camera.

Returns:: true if sensor type is monocular, otherwise false.

inline data_type::SensorData::SensorType get_sensor_type() const

Get the sensor modality associated with this frame.

Returns:: Sensor type value from the underlying sensor data.

inline frame_id_t get_id() const

Get frame identifier.

Returns:: Frame ID.

inline void set_pose(const pose_t &pose_)

Set frame pose in world coordinates.

Parameters:: pose_ – New frame pose.

inline pose_t get_pose() const

Get frame pose in world coordinates.

Returns:: Pose of this frame.

inline data_type::SensorData::weak_ptr get_camera_data() const

Get weak pointer to camera data associated with this frame.

Returns:: Weak pointer to sensor data.

inline void set_number_tracked_points(const int numberTrackedPoints_)

Set number of tracked points matched for this frame.

Parameters:: numberTrackedPoints_ – Number of matched points.

inline int get_number_tracked_points() const

Get number of tracked points matched for this frame.

Returns:: Number of matched points.

inline void update_point_id(size_t index_, point_id_t pointId_)

Used to indicate that a point seen by the frame is in the map already and as map point id equal to pointId (input value).

Parameters:

index_ – Index of the point in frame feature arrays.
pointId_ – Map-point ID corresponding to the indexed frame point.

inline void update_mappoint_id(size_t index_, point_id_t pointId_)

Used to indicate that a point seen by the frame is in the map already and as map point id equal to pointId (input value).

Parameters:

index_ – Index of the point in frame feature arrays.
pointId_ – Map-point ID corresponding to the indexed frame point.

void setup_upgrade_to_keyframe(): Compute the 3D points for each 2D keyfeature.

Note

Assumes RGB-D observations are available to recover depth-backed 3D points for keyframe-upgrade data.

inline const point3d_t::vector &get_3d_points_camera_frame() const

Get 3D points expressed in the camera coordinate frame.

Returns:: Const reference to 3D points in camera frame.

inline const point_id_t::vector &get_mappoint_ids() const

Get IDs of map points associated with this frame.

Returns:: Const reference to map-point IDs.

inline const point2d_t::vector &get_image_points() const

Get 2D image points used in localization for this frame.

Returns:: Const reference to selected image keypoints.

inline const descriptor_t::vector &get_image_descriptors() const

Get descriptors associated with selected frame keypoints.

Returns:: Const reference to descriptor vector.

inline const point2d_t::vector &get_all_detected_image_points() const

Get all detected image keypoints before localization filtering.

Returns:: Const reference to all detected image points.

inline const descriptor_t::vector &get_all_detected_image_descriptors() const

Get descriptors for all detected image keypoints.

Returns:: Const reference to all detected image descriptors.

void get_neighbors(const keyframe_id_t &keyframeId_, std::unordered_map<point_id_t, neighbor_t::vector> &neighbors_) const

Get the neighbors of a point.

Note

This function fills the neighbors variable that is passed by reference, to avoid copying long maps from the caller.

Parameters:

keyframeId_ – Keyframe ID associated with this frame.
neighbors_ – Output map of neighbors for each point.

inline bool check_frame_success()

Check whether frame construction succeeded.

Returns:: true if frame has valid feature content, otherwise false.

void apply_scale_to_points3d(const double scale_)

Apply a uniform scale factor to frame 3D points.

Parameters:: scale_ – Scale factor to apply.

inline void set_reference_keyframe_id(const keyframe_id_t keyframeId_)

Sets the ID of the reference keyframe of this frame. Reference keyframe is the keyframe to which this frame has the most points in common.

Parameters:: keyframeId_ – Reference keyframe ID.

inline keyframe_id_t get_reference_keyframe_id() const

Get the ID of the reference keyframe of this frame.

Returns:: Reference keyframe ID.

inline void set_upgraded_to_keyframe_id(const keyframe_id_t keyframeId_)

Sets the ID of the keyframe this frame is upgraded to.

Parameters:: keyframeId_ – Keyframe ID created from this frame.

inline keyframe_id_t get_upgraded_to_keyframe_id() const

Gets the ID of the keyframe this frame is upgraded to.

Returns:: Keyframe ID this frame was upgraded to.

inline bool is_keyframe() const

Checks if the frame was upgraded to a keyframe.

Returns:: True if the frame was upgraded to a keyframe, False otherwise.

inline bool valid_keyframe_upgrade() const

Check if the frame has the necessary data to be a keyframe.

Returns:: True if the frame has the necessary data to be a keyframe, False otherwise.

Public Static Functions

static inline Frame::shared_ptr make_shared(const data_type::SensorData::weak_ptr cameraData_)

Create a shared Frame instance from sensor data.

Parameters:: cameraData_ – Weak pointer to sensor data associated with the frame.
Returns:: Shared pointer to the created Frame object.

Private Functions

inline bool __check_frame_success()

Check whether detected features meet minimum frame requirements.

Returns:: true if frame has enough detected points, otherwise false.

Private Members

frame_id_t _id: Frame Id.

pose_t _pose = pose_t::Identity(): Frame Pose in world coordinates.

data_type::SensorData::shared_ptr _cameraData: camera data (can be stereo, monocular, or RGB-D from different datasets)

int _numberTrackedPoints = 0: To save the number of keypoints successfully matches with a previous frame/keyframe/buffer (useful for keyframe-creation decisions)

double _minDepth = 0.1: Minimum depth for the 3D points.

double _maxDepth = 10.0: Maximum depth for the 3D points.

bool _createdWithSuccess = true: Flag that indicate if the frame was successfully created.

keyframe_id_t _referenceKeyframe = keyframe_id_t::UNSET: Id of the reference keyframe.

keyframe_id_t _upgradedToKeyframeId = keyframe_id_t::UNSET: Note that not all frames are upgraded.

point2d_t::vector _allDetectedImagePoints: Keep all the detected image points for better accuracy latter.

descriptor_t::vector _allDetectedImageDescriptors: Image Descriptors.

point_id_t::vector _allDetectedImagePointIds: Ids of Mappoints for the image points that are matched to the map.

point3d_t::vector _pointsCameraFrame: 3D points seen by the frame in its coordinate frame

point2d_t::vector _pointsImage: 2D image points

descriptor_t::vector _pointsImageDescriptors: Descriptors of the 2D/3D points.

point_id_t::vector _pointsInMapIds: Ids of all the detected mappoints in the image.