Image Decomposition using Matrix Pencil

D. Haddad and Y. Luzanov (USA)


Matrix pencil, image analysis, image representation


One of the most common approaches of representing im ages in frequency domain, as a sum of complex exponen tials, is by using DFT. But, DFT is limited in the resolu tion of the frequencies that can be computed. The discrete nature of the frequency representation in DFT does not al low for computation of the exact frequencies constituting an image signal. This limitation can be overcome by using matrix pencil (MP) to estimate the exact frequency compo nents of a signal. One of the implications of MP estimation is that a fewer number of parameters is required to repre sent a signal. This paper addresses the application of MP algorithm to estimation of parameters of the sum of com plex exponentials model in 2D images.

Important Links:

Go Back