Cosine-modulated two-dimensional time-varying filter banks

Two-dimensional (2-D) filter banks are widely used for analysis and synthesis systems applied to subband coding of image. Time-varying filter banks that vary the structure of them appropriately for a local property of the signal have been studied for image data compression. In time-varying filter banks, each different part of the signal is processed with different filter coefficients or a different number of channels (a different sampling matrix). Those signals have to be perfectly reconstructed even across the transition of the filter banks. We consider time-varying systems of cosine-modulated 2-D filter banks for arbitrary sampling lattices. We show perfect reconstruction (PR) conditions for such time-varying filter banks that vary the filter coefficients or a sampling matrix. Some applications are presented to show the effectiveness of this method.