ABSTRACT
Once a transfer function that satisfies the requirements imposed by the application at hand is obtained, the implementation of the filter must be undertaken. In a software implementation, the difference equation, a signal-flow graph, a digital-filter network, or a state-space representation is converted into a computer program that can be run on a general-purpose computer or workstation or on a general-purpose digital signal processing chip. In a hardware implementation, the digital filter is implemented in terms of a collection of specialized very-large-scale integrated circuit chips. Although a software implementation is entirely satisfactory for many applications, the highest processing rate that can be achieved is limited by the use of general-purpose hardware. Consequently, in applications where a software implementation cannot provide the required processing rate, an implementation in terms of specialized hardware should be considered. An alternative approach to the implementation of recursive 2-D filters is to use a systolic implementation for 1-D recursive filters as a building block.
