Visual Computing and Image Processing Lab
Oklahoma State University

Imaging, Processing, Inferencing and Learning



Target Tracking and Recognition

We studied the problem of an integrated target detection, tracking, recognition and learning for infrared imagery. Specifically, our research focuses on three interrelated and complementary research problems: appearance representation, dynamic modeling, and algorithm integration. Three issues are addressed. (1) How to provide a general appearance representation for different targets and under different poses. (2) How to develop a powerful dynamic model for maneuvering target tracking that is able to provide an explicit dynamics analysis on-the-fly. (3) How to integrate multiple cues (motion/appearance) and multiple tasks (detection, tracking, recognition and learning) into one Bayesian estimation framework.

(Target synthesis results of the appearance generative model learned from 24 templates for each target for five targets from left to right, the original 3D model, the silhouettes, the distance transform of original silhouettes, the synthesized distance transform/target silhouettes.)

(Target tracking results on four AMCOM infrared closure sequences, where the black and white gaits show the ground truth and the estimated one respectively. Courtesy of the Center of Imaging Science at Johns Hopkins University.)


Copyright © 2008 VCIPL@OSU, All rights reserved.
(Acknowledgements: The template is from Interspire Free Templates, and free pictures are from