Invariant Filtering of Kinematics and Range Signals for Object Tracking and Identification

L. Hong and K. Xue (USA)


Multiple model filtering, filter banks, signal processing, invariants, estimation and identification.


The paper presents an invariant-based filtering algo rithm for moving object tracking and identification us ing kinematics and high resolution range measurements. The algorithm effectively exploits coupled information between object kinematics and identification spaces by introducing the concept of local and global motion. A geometrical invariant constraint based on the object rigidity principle is built into object kinematics and measurement models, which nicely facilitate joint infor mation exploitation. An interacting multiple template (IMT) filtering algorithm is developed for joint tracking and identification. Besides providing object kinematics behavior and identity information, the algorithm is ca pable of reconstructing the prominent physical structure of a moving object.

Important Links:

Go Back