A method for improving tracking in crowded situations using rival compensation
- 32 Abstract A METHOD FOR IMPROVING TRACKING IN CROWDED SITUATIONS USING RIVAL COMPENSATION A method for tracking objects across a number of image frames (100,140), includes tracking objects in the frames based on appearance models (200,240,250) of each foreground region corresponding to each of th...
Gespeichert in:
1. Verfasser: | |
---|---|
Format: | Patent |
Sprache: | eng |
Schlagworte: | |
Online-Zugang: | Volltext bestellen |
Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
Zusammenfassung: | - 32 Abstract A METHOD FOR IMPROVING TRACKING IN CROWDED SITUATIONS USING RIVAL COMPENSATION A method for tracking objects across a number of image frames (100,140), includes tracking objects in the frames based on appearance models (200,240,250) of each foreground region corresponding to each of the objects (130,120,110) and determining if a plurality of the tracked objects overlap (340,390). Where a plurality of the tracked objects overlap, the method creates compensated appearance models (500) for each of the plurality of overlapping objects by attenuating common appearance features (520,530,540) among the corresponding appearance models; and tracks the plurality of overlapping objects based on (500) the created compensated appearance models. 7Q99R1q 1 PnA.qnq cnpni |nrinp 900 Start Initialise track list Get video frame 925 Make track prediction for each track Get video objects 930 Associate video objects with tracks Deal with un-associated tracks and multiply associated objects More video Yes frames? No99 End Fig. 9 |
---|