Translator Disclaimer
4 August 2009 Robust infrared targets tracking with covariance matrix representation
Author Affiliations +
Robust infrared target tracking is an important and challenging research topic in many military and security applications, such as infrared imaging guidance, infrared reconnaissance, scene surveillance, etc. To effectively tackle the nonlinear and non-Gaussian state estimation problems, particle filtering is introduced to construct the theory framework of infrared target tracking. Under this framework, the observation probabilistic model is one of main factors for infrared targets tracking performance. In order to improve the tracking performance, covariance matrices are introduced to represent infrared targets with the multi-features. The observation probabilistic model can be constructed by computing the distance between the reference target's and the target samples' covariance matrix. Because the covariance matrix provides a natural tool for integrating multiple features, and is scale and illumination independent, target representation with covariance matrices can hold strong discriminating ability and robustness. Two experimental results demonstrate the proposed method is effective and robust for different infrared target tracking, such as the sensor ego-motion scene, and the sea-clutter scene.
© (2009) COPYRIGHT Society of Photo-Optical Instrumentation Engineers (SPIE). Downloading of the abstract is permitted for personal use only.
Jian Cheng "Robust infrared targets tracking with covariance matrix representation", Proc. SPIE 7383, International Symposium on Photoelectronic Detection and Imaging 2009: Advances in Infrared Imaging and Applications, 73833H (4 August 2009);


Back to Top