fScan Reference Manual, Chapter 2 (CmdLineOptions): despike
Despike -- Filter extreme values (i.e., shot noise)
usage: -despike Nstdevs flg winsiz
Extreme data values (spikes) can be filtered from the data either by time course or by image or raw data matrix. The default is to filter by time course unless running in real-time mode (-wait flag set). The algorithm first scans the data to determine the range of values that includes most of the data and calculates the standard deviation over that range. In the 2nd pass any value that is greater than NSTDEV's different than the average 4 preceding values is replaced.
NSTDEV - NSTDEV defines the range beyond which filtering is done; default = 3.
FLG - FLG is a bit mask with the following meaning:
1 - filter spikes out of 1-D (time course) data
2 - filter spikes out of 2-D (image or matrix) data
4 - replace spikes with previous good value, otherwise replace with mean of 4 previous values
WINSIZ - WINSIZ is a parameter that controls how much bunching is expected in 1st pass. The default is 4, meaning that in a 8 bin histogram of the distribution of data values, the cut off for the central peak is a any bin with less than WINSIZ percent (e.g. 4%) of the total. Increasing WINSIZ allows more data variability and does less filtering.