SaDVIO
|
Implements feature tracking between two images. More...
#include <afeaturetracker.h>
Public Member Functions | |
AFeatureTracker () | |
AFeatureTracker (std::shared_ptr< AFeatureDetector > detector) | |
virtual uint | track (std::shared_ptr< isae::ImageSensor > &sensor1, std::shared_ptr< isae::ImageSensor > &sensor2, std::vector< std::shared_ptr< AFeature >> &features_to_track, std::vector< std::shared_ptr< AFeature >> &features_init, vec_match &tracks, vec_match &tracks_with_ldmk, int search_width=21, int search_height=21, int nlvls_pyramids=3, double max_err=10, bool backward=false)=0 |
Track features between two sensors. More... | |
Protected Attributes | |
std::shared_ptr< AFeatureDetector > | _detector |
feature detector for feature init More... | |
std::string | _feature_label |
label for the features More... | |
Implements feature tracking between two images.
Contrary to feature matching, feature tracking doesn't require a set of features on the second image. It uses optical flow to estimate the motion of features from the first image to the second image.
|
inline |
|
inline |
|
pure virtual |
Track features between two sensors.
sensor1 | First sensor containing the features to track |
sensor2 | Second sensor where the features will be tracked |
features_to_track | Vector of features to track from sensor1 |
features_init | Vector of features initialized in sensor1 |
tracks | Output vector to store matched features |
tracks_with_ldmk | Output vector to store matched features with landmarks |
search_width | Width of the search area for tracking (default is 21) |
search_height | Height of the search area for tracking (default is 21) |
nlvls_pyramids | Number of pyramid levels for tracking (default is 3) |
max_err | Maximum error for tracking (default is 10) |
backward | Whether to track features backward to filter outliers (default is false) |
Implemented in isae::Line2DFeatureTracker, isae::semanticBBoxFeatureTracker, and isae::Point2DFeatureTracker.
|
protected |
feature detector for feature init
|
protected |
label for the features