Frequency-Domain Separable Decomposition of 2-Dimensional Systems
Fan filters in seismic data processing, directional filters in image processing and 2-D equalizers are but a few examples of 2-Dimensional systems described explicitly in the frequency domain. Current methods for handling such systems in a discrete form require an NxN sampling grid resulting in considerable design complexity and a high implementation cost.
The analysis and design problem is reduced to a 1-D one.
Application of parallel-processing techniques is allowed.
The number of elements required for the implementation is now 2xLxN; in the case where L << N, appreciable simplification and economy is obtained.
Applications are given through design examples of 2-D low-pass, fan and Wiener filtering.
KeywordsImpulse Response Singular Value Decomposition Frequency Response Function Wiener Filter Effective Domain
Unable to display preview. Download preview PDF.