Back to the main page.

Bug 1561 - bandpass filter seems to be direction sensitive

Status CLOSED FIXED
Reported 2012-06-27 14:56:00 +0200
Modified 2012-08-23 10:33:52 +0200
Product: FieldTrip
Component: preproc
Version: unspecified
Hardware: PC
Operating System: Mac OS
Importance: P3 normal
Assigned to: Robert Oostenveld
URL:
Tags:
Depends on:
Blocks:
See also: http://bugzilla.fcdonders.nl/show_bug.cgi?id=1318

Robert Oostenveld - 2012-06-27 14:56:26 +0200

this was reported to me by one of the Hamburg course participants (working on TMS-EEG data).


Jörn M. Horschig - 2012-06-27 16:18:11 +0200

The problem becomes strongly visible when analyzing a signal with a spike/dirac pulse at the last sample. The sample will be 'pushed' out of the signal by the forward filter, while not being pushed back in. As a solution we implemented a two-pass reverse and a two-pass average solution. Two-pass reverse first applies the reverse filter, then the forward filter. Two-pass average will apply both twopass and twopass-reverse and then take the average, i.e. it edge effects will be averaged. Averaged should really be taken in a literal sense, meaning that we will get a /\ like effect. See ft_testbug1561 for this. Also see bug 1318 which deals with implementing more padding options (pad the mirrored signal). 551 $ svn ci -m "bugfix-#1561- added twoway-reverse and twoway-average as additional filter directions" Sending preproc/ft_preproc_bandpassfilter.m Sending preproc/ft_preproc_bandstopfilter.m Sending preproc/ft_preproc_highpassfilter.m Sending preproc/ft_preproc_lowpassfilter.m Sending preproc/private/filter_with_correction.m Transmitting file data ..... Committed revision 6160.


Robert Oostenveld - 2012-08-23 10:33:52 +0200

closed multiple bugs that have been resolved for some time