2-D LMA Filters -Design of Stable Two-Dimensional Digital Filters with Arbitrary Magnitude Function-
スポンサーリンク
概要
- 論文の詳細を見る
This paper proposes a technique for designing two-dimensional (2-D) digital filters approximating an arbitrary magnitude function. The technique is based on 2-D spectral factorization and rational approximation of the complex exponential function. A 2-D spectral factorization technique is used to obtain a recursively computable and stable system with nonsymmetric half-plane support from the desired 2-D magnitude function. Since the obtained system has an exponential function type transfer function and cannot be realized directly in a rational form, a class of realizable 2-D digital filters is introduced to approximate the exponential type transfer function. This class of filters referred to as two-dimensional log magnitude approximation (2-D LMA) filters can be viewed as an extension of the class of 1-D LMA filters to the 2-D case. Filter coefficients are given by the 2-D complex cepstrum coefficients, i.e., the inverse Fourier transform of the logarithm of the given magnitude function, which can be efficiently computed using 2-D FFT algorithm. Consequently, computation of the filter coefficients is straightforward and efficient. A simple stability condition for the 2-D LMA filters is given. Under this condition, the stability of the designed filter is guaranteed. Parallel implementation of the 2-D LMA filters is also discussed. Several examples are presented to demonstrate the design capability.
- Institute of Electronics, Information and Communication Engineersの論文
- 1992-02-20
Institute of Electronics, Information and Communication Engineers | 論文
- Proposal of a Desk-Side Supercomputer with Reconfigurable Data-Paths Using Rapid Single-Flux-Quantum Circuits
- Interacting Self-Timed Pipelines and Elementary Coupling Control Modules
- A Lexicon-Driven Handwritten City-Name Recognition Scheme for Indian Postal Automation
- A Dual Transformation Approach to Current-Mode Filter Synthesis
- Acoustic Feature Transformation Combining Average and Maximum Classification Error Minimization Criteria