vobj_tracker Class Reference
[Object level tracking]

#include <vobj_tracker.h>

Inheritance diagram for vobj_tracker:

Inheritance graph
[legend]
Collaboration diagram for vobj_tracker:

Collaboration graph
[legend]

List of all members.

Public Member Functions

 vobj_tracker (int width, int height, int levels, int max_motion, visual_database *vdb=0, bool glContext=false, pyr_keypoint::pyr_keypoint_factory_t *kf=0, vobj_frame::vobj_frame_factory_t *ff=0, pyr_track::pyr_track_factory_t *tf=0)
virtual pyr_frameprocess_frame (IplImage *im)
 Store a new frame, detect features, and match them with previous frame.
virtual pyr_frameprocess_frame_pipeline (IplImage *im)
 Pipelined version of process_frame.
int track_objects (vobj_frame *frame, vobj_frame *last_frame)
void remove_visible_objects_from_db (vobj_frame *frame)
void incremental_learning (vobj_frame *frame, int track_length, float radius, int max_pts)

Public Attributes

visual_databasevdb
std::auto_ptr< incremental_queryquery
float score_threshold
int max_results
int homography_corresp_threshold
 Threshold to determine successful detection for homography. Default: 20.
int fmat_corresp_threshold
 Threshold to determine successful detection for fundamental matrix. Default: 20.
bool use_incremental_learning

Protected Member Functions

void find_candidates (vobj_frame *frame, std::set< visual_object * > &candidates, vobj_frame *last_frame)
bool verify (vobj_frame *frame, visual_object *obj, vobj_instance *instance)


Detailed Description

Definition at line 90 of file vobj_tracker.h.


Constructor & Destructor Documentation

vobj_tracker::vobj_tracker ( int  width,
int  height,
int  levels,
int  max_motion,
visual_database vdb = 0,
bool  glContext = false,
pyr_keypoint::pyr_keypoint_factory_t kf = 0,
vobj_frame::vobj_frame_factory_t ff = 0,
pyr_track::pyr_track_factory_t tf = 0 
)


Member Function Documentation

void vobj_tracker::find_candidates ( vobj_frame frame,
std::set< visual_object * > &  candidates,
vobj_frame last_frame 
) [protected]

void vobj_tracker::incremental_learning ( vobj_frame frame,
int  track_length,
float  radius,
int  max_pts 
)

pyr_frame * vobj_tracker::process_frame ( IplImage *  im  )  [virtual]

Store a new frame, detect features, and match them with previous frame.

This method takes care of releasing im.

Reimplemented from kpt_tracker.

Definition at line 52 of file vobj_tracker.cpp.

References tracks::get_nth_frame(), incremental_learning(), track_objects(), and use_incremental_learning.

pyr_frame * vobj_tracker::process_frame_pipeline ( IplImage *  im  )  [virtual]

void vobj_tracker::remove_visible_objects_from_db ( vobj_frame frame  ) 

int vobj_tracker::track_objects ( vobj_frame frame,
vobj_frame last_frame 
)

bool vobj_tracker::verify ( vobj_frame frame,
visual_object obj,
vobj_instance instance 
) [protected]


Member Data Documentation

Threshold to determine successful detection for fundamental matrix. Default: 20.

Definition at line 104 of file vobj_tracker.h.

Referenced by verify(), and vobj_tracker().

Threshold to determine successful detection for homography. Default: 20.

Definition at line 101 of file vobj_tracker.h.

Referenced by verify(), and vobj_tracker().

Definition at line 98 of file vobj_tracker.h.

Referenced by find_candidates(), and vobj_tracker().

Definition at line 95 of file vobj_tracker.h.

Referenced by find_candidates().

Definition at line 97 of file vobj_tracker.h.

Referenced by find_candidates(), and vobj_tracker().


The documentation for this class was generated from the following files:

Generated on Tue Mar 30 15:53:26 2010 for Polyora by  doxygen 1.5.9