ملخص
Abstract
In this thesis, we present a method for the design of multidimensional digital filters. This method is based on the use of genetic algorithm (GA). GA algorithm is proposed to optimize the coefficients of magnitude frequency response of digital filter design. GA algorithm is used to minimize a cost function representing the difference between the frequency response magnitudes of an ideal and obtained digital filters. Two types of filters namely, Finite Impulse Response (FIR) and Infinite Impulse Response (IIR) are implemented.The digital FIR filter exhibits a wide transition bands and ripples in the pass band which are considered as disadvantages in digital image processing. Therefore, the only way to obtain a steep transition FIR is to use high order which translates to high cost. An alternative way to avoid high cost filters is to use IIR digital filters. Such digital filters can provide steeper transition bands with lower order than equivalent FIR digital filters. Although FIR digital filters have the disadvantages of high cost in term of hardware realizations, they have the advantage of stability and linear phase. Since IIR digital filters have infinite impulse response, they are prone to unstability. Linear phase which is easy to obtain in FIR digital filters is hard to come by IIR digital filters. The low cost and fast filtering make the IIR digital filters attractive. The proposed GA is utilizing for the design of stable IIR filter with near linear phase constraints. Experimental results are presented for 1-D and 2-D digital IIR filter. From the magnitude-frequency response, the convergence of the proposed approach will be able to obtain global minima in a faster time.