Subspace System Identification of 2-D Stochastic Systems

Description

The fitting of a causal dynamic model to an image is a fundamental problem in image processing, pattern recognition, and computer vision. There are numerous other applications that require a causal dynamic model, such as in scene analysis, machined-parts inspection, and biometric analysis, to name only a few. There are many types of causal dynamic models that have been proposed. Among the most widely known are the autoregressive moving average (ARMA) and state-space models.

In this talk, Ramos will introduce a two-dimensional stochastic state-space system identification algorithm for obtaining stochastic 2-D, causal, recursive, and separable-in-denominator (CRSD) models in the Roesser state-space form. The algorithm is tested with the classical Lena image, and the image reconstructed with the 2-D model is shown to be almost indistinguishable from the true image.

Presenter Bio

Jose Ramos has a Ph.D. and is an Associate Professor at Nova Southeastern University

Date of Event

November 16, 2011 12 - 1:00 PM

Location

Alvin Sherman Library, Room 3015, 3301 College Ave., Fort Lauderdale (main campus)

NSU News Release Link

http://nsunews.nova.edu/mathematics-colloquium-series-lecture-twodimensional-stochastic-systems-nov-16/

Share

COinS
 
Nov 16th, 12:00 PM Nov 16th, 1:00 PM

Subspace System Identification of 2-D Stochastic Systems

Alvin Sherman Library, Room 3015, 3301 College Ave., Fort Lauderdale (main campus)

The fitting of a causal dynamic model to an image is a fundamental problem in image processing, pattern recognition, and computer vision. There are numerous other applications that require a causal dynamic model, such as in scene analysis, machined-parts inspection, and biometric analysis, to name only a few. There are many types of causal dynamic models that have been proposed. Among the most widely known are the autoregressive moving average (ARMA) and state-space models.

In this talk, Ramos will introduce a two-dimensional stochastic state-space system identification algorithm for obtaining stochastic 2-D, causal, recursive, and separable-in-denominator (CRSD) models in the Roesser state-space form. The algorithm is tested with the classical Lena image, and the image reconstructed with the 2-D model is shown to be almost indistinguishable from the true image.

https://nsuworks.nova.edu/mathematics_colloquium/ay_2011-2012/events/10