High Throughput Rate 2-D Fir Filter Design Using Singular Value Decomposition and Genetic Algorithms

This paper presents a method for the design of high throughput rate quadrantally symmetric 2-D filters. It employs cascaded high throughput rate 1-D filters in a parallel structure. These filters require fewer multipliers than a direct 2-D implementation. As well, the parallel structure of the designed filter using SVD lends itself easily to a pipelined, high throughput implementation. The 1-D filters achieve a high throughput rate using canonical signed digit coefficients which are obtained using a new genetic algorithm design method. This new design method employs a unique chromosome coding technique to overcome the limitations previously encountered in using genetic algorithms for such designs