Scribd
Upload
55 views

Linear Filtering Via DFT

This document discusses using the discrete Fourier transform (DFT) to perform linear filtering via circular convolution. It begins by introducing linear time-invariant (LTI) systems and linear convolution. It then shows an example of linear convolution between input and impulse response sequences. The document explains that circular convolution in the time domain is equivalent to multiplication in the frequency domain. This allows using the DFT and fast Fourier transform (FFT) to efficiently perform linear filtering by zero-padding the input and impulse response sequences before circular convolution. An example calculation demonstrates this approach. The document notes that the frequency domain approach is more computationally efficient than direct time domain circular convolution. It concludes by discussing issues that can arise if the DFT size is smaller than

Uploaded by

hirthik
Copyright
© © All Rights Reserved
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
55 views

Linear Filtering Via DFT

This document discusses using the discrete Fourier transform (DFT) to perform linear filtering via circular convolution. It begins by introducing linear time-invariant (LTI) systems and linear convolution. It then shows an example of linear convolution between input and impulse response sequences. The document explains that circular convolution in the time domain is equivalent to multiplication in the frequency domain. This allows using the DFT and fast Fourier transform (FFT) to efficiently perform linear filtering by zero-padding the input and impulse response sequences before circular convolution. An example calculation demonstrates this approach. The document notes that the frequency domain approach is more computationally efficient than direct time domain circular convolution. It concludes by discussing issues that can arise if the DFT size is smaller than

Uploaded by

hirthik
Copyright
© © All Rights Reserved
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
You are on page 1/ 17

LINEAR FILTERING

via
DFT
LTI System - Output – Linear
Convolution

LTI System
x[n] h[n] y[n]

x[n] = [1 2 ]
h[n] = [1 2 3]

03/16/2021 2
LTI SYSTEM OUTPUT

Linear Filtering - Linear Convolution - Aperiodic Convolution

03/16/2021 3
03/16/2021 4
Exercise: Circular Convolution
• x1[n] = [1 2]
• x2[n] = [1 2 3]

• Obtain the circular convolution of x1[n] and x2[n]

• x3[n] = x1[n] * x2[n]

03/16/2021 5
03/16/2021 6
• Circular convolution is of no use to us ,

if our objective is to determine the output o f a linear


filter to a given input sequence.

03/16/2021 7
• How to perform Linear Convolution via Circular
Convolution ????
• Why this question?

Solution:

Circular convolution in time domain is multiplication in frequency


domain.

Frequency-domain approach based on the DFT , is computationally


efficient,

Due to the existence o f efficient algorithms for computing the DFT


–03/16/2021
Fast Fourier Transform (FFT) 8
03/16/2021 9
So, How to perform Linear Convolution via Circular
Convolution ????
• Input x[n]----------- Length N1 LTI System
h[n]
• Impulse response h[n]--------- Length N2
• Output of LTI system y[n] --------- Length N = N1+N2-1
• x[n] = [1 2]; h[n] = [1 2 3]

N point circular convolution of x(n) with h(n) must be equivalent to the linear
convolution of x(n) with h(n).

In other words, by increasing the length of the sequences x(n) and h(n) to N points
(by appending zeros),

and then circularly convolving the resulting sequences, we obtain the same result as
would have been obtained with linear convolution.

Thus with zero padding, the D F T can be used to perform linear filtering

03/16/2021 10
Exercise: Obtain the output of LTI System
x[n] = [1 2 2 1] ; h[n] = [1 2 3] via Circular
Convolution
• 1. Length of LTI system output: Length x[n] + Length h[n] – 1
• = 4+3-1 = 6
2. With appropriate zero padding, obtain Linearly convolved
output via circular convolution
Modified x[n] = [1 2 2 1 0 0]
Modified h[n] = [1 2 3 0 0 0]
Circular convolution: [1 2 2 1 0 0] [1 2 3 0 0 0]
Linear convolution : [1 2 2 1]* [1 2 3]

03/16/2021 11
Circular convolution: [1 2 2 1 0 0] [1 2 3 0 0 0]
Linear convolution : [1 2 2 1]* [1 2 3]

03/16/2021 12
Frequency domain approach– computationally more efficient than time domain
circular convolution

03/16/2021 13
03/16/2021 14
• What happens if 4 point DFT is computed
instead of 6 point DFT.

03/16/2021 15
03/16/2021 16
FILTERING OF LONG DATA
SEQUENCE

03/16/2021 17

You might also like