【正文】
ective noise filtering operations are performed under all conditions,and (b) number of calculations/replacements is reduced and simplified. This results in a simple hardware realization of the filter. MODIFICATION Consider a 3x3processing window, with P5 as the central pixel,as shown in Figure 1. P1 P2 P3 P4 P5 P6 P7 P8 P9 . A 3x3 processing window The output of MMMF as proposed in [4] is Output (X,Y)= Max(P2,P8)if P5﹥ Max[S] P5 if Min [s]﹤ Max[S] Min(P2,P8) if P5﹤ Max[S] (1) Where S is the set of samples surrounding central pixels except(). S={ P1,P2,P3,P7,P8,P9} (2) The principle invoved in the replacement strategy of Equation (1) is that if P5 is corrupted by noise ,it is better to replace its gray level by P2 or P8 than by using Min[S] or Max[S] .also,due to missing lines error,since P4 and P6 may be lost, they are not considered in Equation(2). The limitation of Equation(1) is that when Min[S] or Max[S] are also corrupted by impulse noise, Min[S] or Max[S] is equal to P5,Equation(1)fails to perform the desired filtering overe this limitation following modifications in the replacement strategy of Equation(1),are proposed. Output (X,Y)= Max(P2,P8)if P5≥ Max[S] P5 if Min [s]< P5< Max[S] Min(P2,P8) if P5≤ Max[S] (3) It has been observed