convolve1d: Finite Impulse Response filter

In FIACH: Retrospective Noise Control for fMRI

Description

This function filters data using specified FIR filter coefficients.

Usage

convolve1d(x,fir,subtractMed=TRUE) 
            

Arguments

x	A numeric matrix where the columns are to be filtered.
fir	The FIR filter coefficients. This should always be a sinc or windowed sinc. Using other fir coefficients will lead to wrong results.
subtractMed	Should the Median be subtracted prior to convolution to prevent Gibb's ringing.

Value

Returns the filtered signal.

Examples

## 1.  construct a sinusoid oscillating at .03Hz for 600s with a TR of 2.16s  
## 2. add a sinusoid oscillating at .15Hz for 600s with a TR of 2.16s
## 3. make large matrix of signals

test<-sin(2*pi*.03*(seq(0,600,2.16)))                                   
test.noise<-test+sin(2*pi*.15*(seq(0,600,2.16)))                      
test.mat<-matrix(test.noise,nrow=length(test.noise),ncol=5)        

## compute an impulse response. In this case a kaiser window
kais<-kaiserWin(fl=.08,tw=.025,sf=1/2.16,d.sa=70,d.pbr=.1,type="low") 
system.time(filt<-convolve1d(test.mat,kais,subtractMed=TRUE)) 
par(mfrow=c(3,1))                                                    
ts.plot(test,ylim=c(-2,2))                                              
ts.plot(test.mat[,1],ylim=c(-2,2))                                 
ts.plot(filt[,1],ylim=c(-2,2))                        

FIACH documentation built on May 1, 2019, 8:02 p.m.