Computational efficiency is achieved with frequency masking and PI technique instead of IIR filter usage.
This paper presents a design method that consists of a low-order FIR filter realized with a frequency masking technique, and included in the sharpening method.
In this section is suggested frequency masking technique as an initial filter, in sharpening method realization.
All these properties of sharpening method with frequency masking and PI technique are used in section V for narrowband filter realization.
FIR FILTER SHARPENING METHOD AND FREQUENCY MASKING TECHNIQUE
Frequency masking technique is a multirate technique which is appropriate for sharp linear phase filters realizations [5, 6, and 7].
Although the frequency masking technique is a multirate technique, actually there is no sampling rate change through the filter.
In our case, the amplitude change function is realized with FIR filters and frequency masking technique, thus these filters are ideal candidates for the modified PI technique implementation.
So, with choosing M = 2, the characteristic of frequency masking filter, recorded in out 1 (Figure 10) will be:
* Periodic model filter--replace each delay with M * K delays, M is the frequency masking factor, and K is the H(z) order.
Figure 14 shows the amplitude characteristic of filter sharpening structure with amplitude change function F(H) = [3H.sup.2] - [2H.sup.3], and factor M = 2 of frequency masking subsystem filter.